Australas Phys Eng Sci Med DOI 10.1007/s13246-015-0410-1

EPSM 2015 ABSTRACTS

EPSM2015, Engineering and Physical Sciences in Medicine Museum of New Zealand Te Papa Tongarewa, Wellington, 8–12 November 2015, Conference abstracts

 Australasian College of Physical Scientists and Engineers in Medicine 2016

Contents Monday 9th November 0900–1030 Keynote Speaker Session Safety in radiation medicine – Human, hazards and design Todd Pawlicki Cerenkov Luminescence Imaging Simon Cherry 1100–1200 Invited Speaker Session Battling Maxwell’s equations: Physics challenges and solutions for hybrid MRI systems Paul Keall Advanced technology in clinical trials Tomas Kron 1100–1200 Concurrent session Countouring and deformable image registration 1100–1200 Concurrent session Radiobiology modelling 1315–1400 Keynote Speaker Session Molecular imaging for targeted therapies David Jaffray 1400–1430 Invited Speaker Targeted alpha therapy for metastatic melanoma Barry Allen 1400–1530 Concurrent Session Molecular targeted therapies

123

Australas Phys Eng Sci Med

1400–1530 Concurrent Session Stereotactic RT: Intercranial (I) 1400–1430 Concurrent Session Professional issues 1430–1515 Concurrent Session Animal models 1600–1730 Concurrent Session TPS Beam modelling 1600–1645 Concurrent Session Stereotactic RT: Intercranial (II) 1645–1730 Concurrent Session Audits / trials / surveys

1545–1615 Invited Speaker EPID Dosimetry: pre-treatment verification, clinical trial credentialing and real-time patient dose monitoring Peter Greer 1545–1730 Concurrent Session Portal Dosimetry 1545–1615 Invited Speaker Brachytherapy – yesterday, today but is there a tomorrow? Annette Haworth 1545-1730 Concurrent Session Brachytherapy

1600-1730 Concurrent Session Working group women in medical physics

Wednesday 11th November 0800-0900 Concurrent Session

Tuesday 10th November 0900–1030 Keynote Speaker Session Prospective risk analysis in radiation medicine Todd Pawlicki Total-body positron emission tomography Simon Cherry 1100–1130 Invited Speaker Session MARS spectral molecular imaging Anthony Butler 1130–1215 Concurrent Session MRI pre-treatment imaging 1100–1130 Concurrent Session Treatment technique development 1130–1200 Concurrent Session New phantom design 1100–1200 Concurrent Session 3D Printing 1315–1400 Keynote Speaker Session Dosimetry for the MRI linac Bas Raaymakers 1400–1430 Invited Speaker

0900–0945 Keynote Speaker Session The MRI Linac in UMC Utrecht, development and current status Bas Raaymakers 0945–1030 Concurrent Session MRI Linac 0945–1015 Concurrent Session Individual patient QA (I) 0945–1030 Concurrent Session Interventional Radiology 1100–1145 Keynote Speaker Session Image guided radiotherapy David Jaffray 1145–1245 Concurrent Session Real time organ motion tracking and compensation (I) 1145–1230 Concurrent Session Individual patient QA (II) 1145–1245 Invited Speaker Session The role of diagnostic radiology dosimetry in diagnostic reference levels Donald Mclean

Small field dosimetry: still not a small problem

Medical imaging low level radiation and risk

David Thwaites

Marissa Bartlett

1400–1515 Concurrent Session Detectors / Dosimetry (I) 1400–1445 Concurrent Session Pre-treatment imaging 1430–1445 Elekta Travel Award 1400–1515 Concurrent Session Radiology 1545–1715 Concurrent Session Detectors / dosimetry (II)

123

ACDS and ARPANSA audits and surveys

1345–1500 Concurrent Session Real time organ motion tracking and compensation (II) 1345–1500 Concurrent Session Electronic individual patient QA 1345–1500 Concurrent Session Radiology dosimetry (I) 1530–1615 Keynote Speaker Session Statistical process control in radiation medicine Todd Pawlicki

Australas Phys Eng Sci Med

1615–1730 Concurrent Session Patient positioning and organ motion 1615–1715 Concurrent Session

design, which is an open area of research related to safety engineering, will play an important role in determining how to effectively address hazards. The approach of human-centered design will be described using an example from radiation oncology.

Australian synchrotron Thursday 12th November

KS02 Biomedical Applications of Cerenkov Luminescence

0800–0845 Keynote Speaker Session

S. R. Cherry

On-line plan adaptations for MRI guided radiotherapy Bas Raaymakers

Department of Biomedical Engineering, University of California, Davis, USA ([email protected])

0845–0930 International Day of Medical Physics lecture Back to the future: some synergies between physics and medicine from history to horizon David Thwaites 0930–1000 Concurrent Session Global data mining 0930–1000 Concurrent Session Linac QA (I) 1030–1145 Concurrent Session SBRT 1030–1200 Concurrent Session Linac (Q A) II

Poster Abstracts KS01 Safety in radiation medicine – Humans, hazards and design T. Pawlicki

Cerenkov radiation is a phenomenon where optical photons are emitted when a charged particle moves through a dielectric medium faster than the speed of light in that same medium. A number of laboratories have shown that measurable visible light due to the Cerenkov effect is produced in vivo by b-emitting radionuclides commonly used in biomedical research. Furthermore, the amounts of injected activity necessary to produce a detectable signal are consistent with molecular imaging applications. This observation has led to the development of a new hybrid molecular imaging modality known as Cerenkov luminescence imaging (CLI) that allows the spatial distribution of biomolecules labeled with b-emitting radionuclides to be imaged in vivo using sensitive chargecoupled device (CCD) cameras. Two other potential applications of Cerenkov luminescence are for monitoring external beam radiotherapy and as a method to deliver light deep inside tissues to activate phototherapies. This presentation will review the physics of Cerenkov radiation as it relates to in vivo imaging, and present computational results that predict the light yield and spatial distribution for a range of biomedicallyrelevant radionuclides. In vivo studies preclinical imaging studies using both b+ and b–-emitting radionuclides will be to illustrate possible applications of the technique. The use of Cerenkov luminescence for monitoring radiation therapy and for activating phototherapy will also be presented. Finally, the potential for clinical applications will be discussed, and the strengths and weaknesses of Cerenkov imaging compared with other nuclear imaging techniques defined.

Professor and Vice Chair, Dept of Radiation Medicine & Applied Sciences, Univ of California San Diego, USA. ([email protected]) The field of radiation oncology has a long history of quality and safety efforts in both the medical and technical aspects. In spite of these efforts, the field still suffers from routine or frequent small deviations in care as well as infrequent catastrophic errors. With the continuous move towards value based health care systems; the engagement of health care professionals is essential to improve the quality and safety of any health care system. An obstacle in continuing to bring about meaningful quality and safety improvements in the health care setting is the challenge of educating ourselves about quality and safety tools and techniques that have been used successfully in other industries. Similarly, it is not always obvious how best we can adapt those tools and techniques in our domain to realize the benefits. For example, hazard analysis models are not designed to provide the tools to mitigate hazards once the hazards are identified. It is in this area, i.e., methods to create effective improvements, that the largest benefit towards quality and safety improvement may be realized. This presentation will include a discussion of human physical and cognitive biases leading to the understanding that ‘human error’ is never a reason for an accident. The difference between a safety culture and a just culture will be presented leading to the concept of incident learning systems that help identify and understand where our biases and other errors play out in clinical practice. Human-centered

IS01 Battling Maxwell’s equations: Physics challenges and solutions for hybrid MRI systems Paul Keall1, Brendan Whelan1, Brad Oborn2, Stuart Crozier3, Gary Liney4, Lois Holloway4, Michael Barton4 1

Sydney Medical School, University of Sydney and Ingham Institute, Liverpool. ([email protected]). 2Illawarra Cancer Care Centre, Wollongong. 3Department of Information Technology and Electrical Engineering, University of Queensland. 4Ingham Institute, Liverpool Acknowledging the many members of the Australian MRI-Linac Program and funding sources. MRI-guided treatment is a growing area of medicine, particularly in radiotherapy and surgery. The exquisite soft tissue anatomic contrast offered by MRI, along with functional imaging, makes the use of MRI during therapeutic procedures very attractive. Challenging the utility of MRI in the therapy room are many issues including the physics of MRI and the impact on the environment and therapeutic instruments, the impact of the room and instruments on the MRI; safety, space, design and cost. Despite this complexity, integrated MRIguided radiotherapy is now a clinical reality and additional systems are under development. In this talk, the applications and challenges of

123

Australas Phys Eng Sci Med MRIguided radiotherapy in general will be described, with particular emphasis on the solutions developed for the Australian MRI-Linac program.

IS02 Advanced technology in clinical trials T. Kron Peter MacCallum Cancer Centre, Melbourne, Australia. ([email protected]) Introduction Clinical trials are the accepted method to test if patients can benefit from a medical intervention. Advanced technology on one hand relies on clinical trials to demonstrate its impact and utility for patient care, a fact that applies particularly to radiation oncology where technology change has had a huge impact in the last two decades. On the other hand, advanced technology can support clinical trials more generally through the generation of quality data. Methods Radiotherapy clinical trials can be distinguished into three broad categories: 1. Technology is not an important part of the question 2. Technology is required to deliver radiation in a way that is compatible with the trial objectives 3. Technology itself is part of the trial question Credentialing of centres participating in trials of category 2 and 3 has become standard practice for radiotherapy trials as does ‘real time’ review of patient plans. The presentation will provide examples for these activities and review what patients can expect of trials of category 3 Results Stereotactic ablative radiation therapy (SABR) is one of the advanced technologies which is applicable to many tumour sites across curative and palliative cancer treatment. Methods of credentialing include approaches that rely on site visits as well as audits that can be performed remotely. A particular challenge is the credentialing of image guidance procedures that are essential for the conformal radiation delivery afforded by SABR. This will be discussed in the context of a completed (BOLART) and an upcoming (RAIDER B) trial for adaptive radiotherapy for bladder cancer. Conclusion Advanced technology is an integral part of many clinical trials. It can benefit from clinical trials that can demonstrate appropriate indication and the utility for a technology as well as adding value to trials by recording accurate delivery and possibly indicating clinical outcomes early. Acknowledgements The author acknowledges many discussions with members of the QA teams of TROG, IROC and RTTQA.

O001 Atlas based auto-segmentation method incorporating delineation uncertainty for whole breast radiotherapy L. R. Bell1,2, J. Dowling3, E. M. Pogson1,2, P. Metcalfe1,2, L. Holloway1,2,4,5 1 Centre for Medical Radiation Physics, University of Wollongong, Australia. 2Liverpool & Macarthur Cancer Therapy Centres & Ingham Institute, Liverpool, Australia. ([email protected]), ([email protected]), ([email protected]). 3Australian e-Health Research Centre, CSIRO, Queensland, Australia. ([email protected]). 4SWSCS, University of New South Wales, Australia. 5Institute of Medical Physics, University of Sydney, Australia. ([email protected])

123

Introduction Adaptive radiotherapy (RT) delivered with highly conformal techniques requires accurate efficient automated contouring methods to minimise the time required and enable effective clinical implementation. This may also enable reduction in inter-observer variation. A method for atlas-based auto-segmentation for whole breast RT that accounts for inter-observer variation is devised and assessed. Methods 28 CT datasets with whole breast CTVs delineated by eight observers were used. The cohort was divided by mean body mass index and laterality into four categories. An average atlas for each category was generated in a leave-one-out approach using the MILXView platform (Dowling, 2012). For each category, observer CTVs were merged to create a contour probability model representing inter-patient and inter-observer differences. The probability model was thresholded to 50 % to generate an autosegmented whole breast CTV contour. Each atlas was registered to the remaining CT and the autosegmented CTV was mapped to this patient, clipped to the patient surface. The STAPLE contour (Warfield, 2004) was generated from the observer CTVs to represent the ‘true’ contour. The dice similarity coefficient (DSC) of the autosegmented CTV with the STAPLE, the smallest and the largest CTV was determined to assess the accuracy of the autosegmentation. Results The DSC results are shown in Table 1. The average time required to autosegment was 3 min, 43 s. Conclusion The atlas-based autosegmentation method for accounting for delineation uncertainty was shown to be fast and accurate with good agreement (DSC [ 0.7) between the autosegmented contours and all CTV volumes. Table 1 DSC results for each volume category Large Left Large Right Small Left Small Right STAPLE

0.81

0.85

0.71

Smallest CTV 0.77

0.86

0.71

0.76

Largest CTV

0.80

0.71

0.79

0.81

0.79

References 1. Dowling, J.A. Lambert, J. Parker, J (2012) 2. Warfield, S.K. Zou, K.H. Wells, W.M. (2004)

O002 Assessing a delineation margin for modern radiotherapy L. R. Bell1,2, J. Dowling3, E. M. Pogson1,2, P. Metcalfe1,2, L. Holloway1,2,4,5 1 Centre for Medical Radiation Physics, University of Wollongong, Australia. 2Liverpool & Macarthur Cancer Therapy Centres & Ingham Institute, Liverpool, Australia. ([email protected]), ([email protected]), ([email protected]). 3Australian e-Health Research Centre, CSIRO, Queensland, Australia. ([email protected]). 4SWSCS, University of New South Wales, Australia. 5Institute of Medical Physics, University of Sydney, Australia. ([email protected])

Introduction Increasingly conformal radiotherapy techniques require accurate contour delineation to ensure we do not miss the cancer. Methods to account for delineation uncertainty must be implemented to minimise any negative impact on dosimetry and patient outcomes

Australas Phys Eng Sci Med

Fig. 1 a Missed tissue (light) and extra tissue (dark). PTV (solid line), CTV union (dashed) and consensus (dotted). b Extra tissue when margin applied to i) Consensus CTV ii) Smallest iii) Largest (Peulen 2015; Pogson 2014). In this work a margin to account for spatially varying delineation uncertainty is assessed. Methods A whole breast patient radiotherapy dataset (n = 21) with eight independent observer CTVs was utilised as a proof of concept. The cohort was divided into categories based on CTV volume and laterality. The margin was defined as twice the standard deviation of contours, averaged for each category across all patients. Margins were determined in both cylindrical and spherical coordinates at angular increments and applied to the smallest, largest and consensus CTV. The overlap of these volumes with observer CTVs was determined. An estimate of the amount of excess normal tissue included and missed tissue was made by determining the volumes in Fig. 1a. Statistical comparison between cylindrical and spherical margins was made. Results The margin encompassment was high with [93 % average overlap of each category’s PTV volume with observer CTVs. Spherically defined margins have significantly greater overlap than cylindrically (p \ 0.05). The average extra tissue (Fig. 1b) was less than 20 % of the CTV union in all cases. There was no missed tissue. No significant differences (p \ 0.05) for extra and missed tissue between spherically and cylindrically defined margins were observed. Conclusion This delineation margin for whole breast radiotherapy has [94 % CTV encompassment, includes less than 20 % extra tissue and misses no tissue when defined in cylindrical and spherical coordinates. Spherically defined margins are more appropriate with superior CTV encompassment with no differences in extra tissue and missed tissue compared to cylindrical. References 1. Peulen, H. Belderbos, J. Guckenberger, M. et al. (2015) 2. Pogson, E. Bell, L. Batumalai, V. et al. (2014)

O003 Evaluation of deformable image registration J. Bird1, S. Marsh2, G. Bengua2, R. Sims3 1

Medical Physics Registrar, ADHB. ([email protected]). 2Director of Medical physics, University of Canterbury. ([email protected]), ([email protected]). 3Medical Physicist, Auckland Radiation Oncology. ([email protected]) Introduction Deformable image registration (DIR) is a type of registration that calculates a deformable vector field (DVF) between two image data sets and permits contour and dose propagation [1]. However the calculation of a DVF is considered an ill-posed problem, as there is no exact solution to a deformation problem, therefore all DVFs calculated contain errors [2]. As a result it is important to evaluate and assess the accuracy and limitations of any DIR algorithm intended for clinical use. Method The hybrid DIR algorithm in RayStation 4.0.1.4 was assessed using a number of evaluation methods and data. The evaluation methods were point of interest propagation, contour propagation and

dose measurements. The data types used were phantom and patient data. A number of metrics were used for quantitative analysis and visual inspection was used for qualitative analysis. Results The quantitative and qualitative results indicated that all DVFs calculated by the DIR algorithm contained errors which translated into errors in the propagated contours and propagated dose. The results showed that the errors were largest for small contour volumes (\20 cm3) and for large anatomical volume changes between the image sets, which pushes the algorithms ability to deform, a significant decrease in accuracy was observed for anatomical volume changes of greater than 10 %. When the propagated contours in the head and neck were used for planning the errors in the DVF were found to cause under dosing to the target tumour by up to 32 % and over dosing to the organs at risk by up to 12 % which is clinically significant. Conclusion The results indicate that contour propagation and dose propagation must be used with caution if clinical use is intended. For clinical use contour propagation requires evaluation of every propagated contour by an expert user and dose propagation requires thorough evaluation of the DVF. References 1. Brock, K.K., Image Processing in Radiation Therapy. 2013: CRC Press. 2. Sotiras, A., C. Davatzikos, and N. Paragios, Deformable medical image registration: A survey. Medical Imaging, IEEE Transactions on, 2013. 32(7): p. 1153–1190.

O004 On the need of systematic strategies for evaluation of deformable image registration used for different purposes A. U. Yeo Physics Department, Radiation Oncology Victoria, GenesisCare, VIC, Australia. ([email protected]) Introduction Various deformable image registration (DIR) algorithms exist in public domain and these were used by numerous studies for different application purposes. DIR performance is often validated based on existing landmarks. What is not well-known is how one can ensure the accuracy of DIR, which can vary depending on its use in different applications. This work suggests various strategies for validating DIR performance for different applications. Methods Various DIR applications are categorised as: (i) image fusion for estimating anatomy changes, (ii) deforming contours, (iii) deforming dose or SUV (from PET) distributions, and (iv) monitoring disease progression. These possess different extents of susceptibility to DIR uncertainty in low-contrast regions. Deformable gel phantom (Yeo et al. 2012) and digitally generated structures were used to evaluate a set of algorithms, demonstrating what factors are more important for different applications of DIR implementation. Results DIR accuracy varies by [200 %, depending on how extensive high-contrast features are used for volume-of-interest (VOI). DIR performance also significantly depends on what parameter sets were used. DIR accuracy of optimised DIR methods tends to be dominated by the precision of landmark identifications. DIR performs better when there is less extent of deformation and vice versa. A good performance of DIR algorithms to match highcontrast features does not necessarily meant such extent of

123

Australas Phys Eng Sci Med performance in low-contrast area, implying that different evaluation strategy needs to be implemented for different DIR applications. Conclusion DIR performance inherently relies on high-contrast features. When DIR is applied to deform high-contrast features and/or regions, the evaluation can be sensibly rely on such highcontrast features. However, when DIR application is associated with deformation and/or registration in low-contrast regions, the evaluation method should not rely on additional or artificial highcontrast features—otherwise the evaluation of DIR accuracy is to be circular. References 1. Yeo, U. A. et al. (2013) Performance of 12 DIR algorithms in lowcontrast regions for mass and density conserving deformation. Med. Phy. 40 (10), 101701

O005 Predicting cancer metastasis via the application of geospatial analysis C. J. Colyer1, 2, M. Bhat1,2 1

School of Physical Sciences, University of Adelaide, Australia. Adelaide Radiotherapy Centre, Adelaide, Australia. ([email protected]), ([email protected])

2

Introduction Cancer is a broad group of diseases that all involve unregulated cell growth. The spread of cancer from the primary site to another site is known as metastasis. The cells in metastases are similar to those in the original tumour and are thus distinguishable from the surrounding tissue. It is well known that certain tumours tend to metastasize in particular organs [1]. The propensity for a particular metastatic cell to spread to a particular organ is termed ‘organotropism’, and the common sites for the different types of primary tumour have been observed clinically. For example, breast cancer frequently metastasises to bone, the brain, the liver and the lungs [2]. The aim of this research is to predict the most probable metastatic sites for an individual with a given primary tumour, which could lead to targeted follow-up programmes, prophylactic treatments or enhanced primary tumour identification when origin is unknown. Method Geospatial analysis is the application of statistical analysis to data which has a geographical aspect. Developed for problems in the environmental and life sciences, geospatial analysis has already been extended to many other fields, including medicine. At the heart of geospatial analysis is the generation of models, such as simple distance decay models or more complex vector models, which can be used to explore correlations between data sets. Correlations between metastases and other data sets, such as the primary tumour, arteries or lymph nodes, will be explored. This will investigate whether metastasis correlates with certain body structures, confirm existing organotropic relationships and attempt to predict metastasis with better success than random chance. Conclusion The results of this study are preliminary. Instead, the focus will be on the methodology involved and its implementation. References 1. Paget S, Cancer Metastasis Rev. 8, 98 (1989). 2. Disibio G, French SW, Arch Pathol Lab Med 132, 931 (2008).

123

O006 Dose–response models of urethral strictures for prostate patients treated with HDR brachytherapy Vanessa Panettieri1, Tiziana Rancati2, Eva Onjukka3, Karen Scott1, Ryan L. Smith1,4. Jeremy L. Millar1,4 1

William Buckland Radiotherapy Centre, The Alfred Hospital, Melbourne, Australia. ([email protected]). 2Prostate Cancer Program, Fondazione IRCCS, Istituto Nazionale dei Tumori, Milan, Italy. 3Dept of Hospital Physics, Karolinska University Hospital, Stockholm, Sweden. 4School of Applied Sciences, RMIT University, Melbourne, Australia Introduction In the treatment of prostate cancer the use of HDR brachytherapy in combination with external beam radiotherapy (EBRT) is an established technique which provides biochemical tumour control comparable to EBRT-only while minimising doses to the organs at risk. Despite being widely used there is no consensus on the optimal fractionation regime (Hoskin 2013) making it difficult to establish dose limits, in particular for the urethra. Our aim was to assess best dosimetric predictors of urethral strictures and to create dose–response models of such toxicity. Methods A cohort of 281 patients treated in our institution was retrospectively analysed. The patients had follow-up 6, 12, 18, 24 months and then every year until 10 years after the treatment. Urethral DVHs and clinical factors were exported from our database, and univariate/multivariate logistic regressions were used to estimate best dosimetric predictors of urethral stricture. Since the patients were treated using different fractionations regimes (18 Gy in 3, 19 Gy in 2 and 17 Gy in 2 fractions) physical doses were converted into BED (a/b = 5 Gy). By using the regression model dose–response curves were created. The endpoint considered was the time of the first urethrotomy; the median follow-up time was 7 years, and the overall stricture rate was 15 %. Results The univariate analysis showed that dosimetric parameters such as BED10 % and BED5 % were significantly correlated to urethra stricture (p = 0.001, OR = 1.05, AUC = 0.63–0.64). For BED10 % the dose response curve (Fig. 1a) was representative of observed toxicity rates for BED [ 50 Gy. Among clinical factors the use of neoadjuvant androgen deprivation showed protective effect and improved the logistic model (OR = 0.46, AUC = 0.68) (Fig. 1b). Conclusion Doses to small volumes of the urethra are significantly correlated to urethra stricture. Calculated dose–response curves were representative of the toxicity rate and showed that androgen deprivation acts as a dose modifier, proving the importance of adding clinical factors in the models to predict toxicity. References 1. Hoskin PJ, Colombo A, Henry A, Niehoff P, Hellebust TP, Siebert FA, Kovacs G GEC/ESTRO reccomandations on high dose rate afterloading brachytherapy for localised prostate cancer: An update, Radioth. Oncol. 107 (2013) 325–332.

Fig. 1 Dose-response curves modelling urethral stricture probability as a function of BED10 %. a shows the result of the univariate analysis and b the results when Androgen deprivation (AD) is added in the multivariate analysis

Australas Phys Eng Sci Med

O007 Parameters for the Lyman–Kutcher–Burman model of normal tissue complication probability based on late rectal bleeding events after prostate radiotherapy

[2–4]. The n (TD50) decreased (increased) as the criteria for rectal bleeding increased. Consequently, the response became more serial (i.e. more dependent on the highest doses) as the severity of rectal bleeding increased. Including a correction for dose fractionation did not significantly improve the maximum likelihood.

C. R. Moulton1, M. House1, V. Lye2, C. Tang2, M. Krawiec2, A. Kennedy2, D. J. Joseph2, J. Denham3, M. A. Ebert1,2

References

1

School of Physics, University of Western Australia, Crawley, Western Australia. ([email protected]), ([email protected]). 2Radiation Oncology, Sir Charles Gairdner Hospital, Nedlands, Western Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]). 3School of Medicine and Population Health, University of Newcastle, New South Wales, Australia. ([email protected]) Introduction This study calculates the Normal Tissue Complication Probability (NTCP) parameter values according to the LymanKutcher-Burman (LKB) model [1] for late rectal bleeding. Methods In the RADAR trial, 724 patients received 66, 70 or 74 Gy of external beam radiotherapy (EBRT) and 167 patients received 46 GyGy of EBRT and 19.5 Gy of high-dose-rate (HDR) brachytherapy. For patients receiving EBRT only, the dose-volume parameters were obtained. For the combined treatment: (i) the EBRT CT was registered to the HDR CT with a rigid + scale + deformable registration in Velocity AI and (ii) the dose-volume parameters were obtained from the unregistered EBRT and registered HDR Acuros dose distributions. Patients were classed into toxicity or no toxicity groups if they did or did not have at least a certain grade LENTSOMA late rectal bleed. For each patient, the NTCP [1] was calculated for every combination in the parameter space (m = 0:0.01:0.6, n = 0.02:0.01:0.8, TD50 = 33:1:137). The log-likelihood [1] was calculated for each parameter combination and the maximum loglikelihood estimates were determined. The 95 % confidence interval (CI) for each parameter was calculated via the profile likelihood method [1]. Results The table provides the maximum log-likelihood estimates and 95 % CIs. The impact of dose fractionation was considered by converting to equieffective doses at 2 Gy/fraction [2] (a/ b = 0.1:0.1:10); however, this did not significantly improve the maximum log-likelihood (p [ 0.05, likelihood ratio test [2]).

Toxicity

# tox / # no tox m (95 % CI)

n (95 % CI)

TD50 (95% CI)

Patients who received EBRT only C2

183/541

0.28 (0.18–0.54) 0.16 (0.06–0.42) 69 (64–82)

C3

66/658

0.20 (0.13–0.04) 0.13 (0.05–0.41) 79 (72–115)

All patients C2

0.38 (0.26–0.60)

C3

0.21 (0.16–0.33)

0.119 (0.10–0.46) 0.19 (0.10–0.46) 73 (66–88) 0.14 (0.09–0.28) 0.14 (0.09–0.28) 80 (73–103)

VIMC and Acuros doses are dose to water values

Conclusion The 95 % CIs for the n and TD50 parameters covered the range of estimates from studies [2–4]. However, the estimates of the slope of the dose–response curve (m) were larger than other studies

1. Carolan, M. et al. 2014. J. Phys.: Conf. Ser., 489, 012087. 2. Tucker, S. et al. 2011. Int. J. Radiat. Oncol. Biol. Phys., 81, 600–605. 3. Gulliford, S. et al. 2012. Radiother. Oncol., 102, 347–351. 4. Michalski, J. et al. 2010. Int. J. Radiat. Oncol. Biol. Phys., 76, S123-S129.

O008 The impact of statistical modelling strategies on the performance of predictive models for urinary symptoms following external beam radiotherapy of the prostate N. Yahya1, M. A. Ebert1,2, M. Bulsara3, M. J. House1, A. Kennedy2, D. J. Joseph2,4, J. W. Denham5 1

School of Physics, University of Western Australia, Western Australia, Australia ([email protected]). 2 Department of Radiation Oncology, Sir Charles Gairdner Hospital, Western Australia, Australia. 3Institute for Health Research, University of Notre Dame, Fremantle, Western Australia, Australia. 4 School of Surgery, University of Western Australia, Western Australia, Australia. 5School of Medicine and Public Health, University of Newcastle, New South Wales, Australia Introduction In the era of dose-escalated radiotherapy for prostate cancer, urinary toxicity will likely represent the principal dose-limiting factor. Given the paucity of available urinary toxicity data, it is important to ensure derivation of the most robust models. New algorithms are commonly used in place of simpler methods due to the potentially superior predictive performance. This work explores different statistical modelling strategies for urinary symptoms and compares the predictive performance. Methods The performance of logistic regression (LR), elastic-net, support-vector machine (SVM), random forest (RF), neural network (NN) and enhanced adaptive regression through hinges (EARTH) to predict urinary symptoms were analysed using data from 754 participants accrued by TROG03.04-RADAR. Dysuria, haematuria, incontinence and frequency, with three definitions (grade C 1, grade C 2 and longitudinal) with event rate between 2.3–76.1 % were analysed. Repeated 10-fold cross-validations were performed. Parameters were optimised on training data. Area under the receiver operating curve (AUROC) were calculated and compared. Results RF, LR, elastic-net and EARTH were the highest performing in 5, 3, 2 and 2 endpoints. In 7 endpoints, the differences were not statistically significant to one or more other methods. RF, LR, elasticnet, EARTH, SVM and NN were the best/not significantly worse in 7, 7, 6, 6, 1 and 1 endpoints. The AUROC difference between the best and worst algorithms were largest (0.18, 0.18, 0.16) for endpoints with low events (2.3 %, 7.1 %, 3.5 %). Average best-worst difference for others was 0.06. The best AUROC was 0.662. Conclusion RF was most likely to be the best performing algorithm with LR and elastic-net producing competitive results. Careful selection of an algorithm is crucial in instances where the event rate is low. The predictive power of the models was modest, and new

123

Australas Phys Eng Sci Med features, including spatial dose maps, are required for better predictive capability.

KS03 Molecular imaging for targeted therapies D. Jaffray

6. Huang CY, et al. (2012) Monte Carlo calculation of the maximum therapeutic gain of Tumor Anti-vascular Alpha Therapy. Medical Physics 39(3), 1282–1288 7. Song EY, et al. (2008) The cytokinesis–block assay as a biological dosimeter for targeted alpha therapy. Phys Med Biol. 53, 319–328 8. Allen, BJ, et al. (2013) Alpha emitting radionuclides and radiopharmaceuticals for therapy, IAEA, Vienna, http://www-naweb. iaea.org/napc/iachem/working_materials.html

Director, Institute of Health Technology Development, University Health Network (TECHNA), Canada Abstract not yet supplied

O009 Covalently conjugated paclitaxel to gold nanoparticles for breast cancer treatment Z. Alhalili1, D. Figueroa2, B. Sanderson2, J. Shapter1

Barry J. Allen Faculty of Medicine, University of Western Sydney ([email protected]) Introduction External beam radiotherapy plays a major role for early curative or late palliative therapy. However, an important issue not being addressed is systemic therapy for systemic disease. As such, I began a R&D program for targeted alpha therapy (TAT), which led to a world first, phase 1 clinical trial of TAT for metastatic melanoma using the Bi213-9.2.27 alpha immunoconjugate (AIC). Methods This AIC was tested in vitro and in melanoma bearing nude mice followed by phase 1 trials of intralesional (N = 15) and systemic (N = 45) therapy for melanoma. Results The preclinical studies showed efficacy without adverse events. The phase 1 trial of intralesional therapy1 demonstrated efficacy without adverse events. The subsequent trial of systemic (intravenous) therapy2 gave tumour regression in 10 subjects and half the patients had stable disease, all without adverse events.3 Discussion The observed efficacy was surprising in that the short range of alpha radiation would inhibit tumour regression. The tumour anti-vascular alpha therapy (TAVAT) concept5 was proposed to account for this effect and was later validated by microscopic Monte Carlo calculations6. Current dosimetry is not applicable for high LET radiation. A world first study7of a biological dosimeter using the micronucleus technique was demonstrated in the melanoma clinical trial. The IAEA has reported on the current status and future directions for this therapy8. A comparison of TAT with current therapeutic modalities4 shows that TAT still has an application today. But are there any medical physicists with the multidisciplinary skills and opportunity to continue the development and demonstration of systemic alpha therapy for metastatic cancer? References 1. Allen BJ, et al. (2005) Intralesional targeted alpha therapy for metastatic melanoma. Can Biol Therapy, 4 (12); 1318–1324 2. Raja C, et al. (2007) Interim analysis of toxicity and response in Phase 1 trial of systemic targeted alpha therapy for metastatic melanoma. Can Biol Therapy. 6: 846–52 3. Allen BJ, et al. (2011) Analysis of Patient Survival in a Phase 1 Trial of Systemic Targeted Alpha Therapy for Metastatic Melanoma. Immmunotherapy, 3(9), 1041–1050. 4. Brown MP, et al. (2015) The potential complementary role of targeted alpha therapy (TAT) in the management of metastatic melanoma. Melanoma Management. in press 5. Allen BJ, et al. (2007) Tumour anti-vascular alpha therapy: a mechanism for the regression of solid tumours in metastatic cancer. Phys. Med. Biol. 52 L15-L19.

123

1

Flinders Centre for NanoScale Science and Technology, School of Chemical and Physical Sciences, Flinders University, Adelaide, Australia. ([email protected]), ([email protected]). 2 School of Medical Science and Technology, Flinders University, Adelaide, Australia. ([email protected]), ([email protected]) Introduction Paclitaxel is an anticancer drug (Li, 2014). However, it has limitations involve non-specific targeting of the malignant cells (Syed, 2013), and insufficient dosages reaching the tumor due to the non-specific distribution which leads to severe side effects (Wang, 2008). Therefore, nanotechnology has emerged with novel materials exhibit unique physical, chemical, and biological properties (Ochekpe, 2009) (Lee, 2014). Among them, gold nanoparticles, AuNPs, have attracted significant attention due to their unique properties such as high surface area, small size, biocompatibility, noncytotoxicity properties making them promising in medical applications (Kumar, 2012) (Pissuwan, 2011) (Murawala, 2014). The aim of this study was to develop a novel drug delivery system combines AuNPs with paclitaxel to improve the limitations exist in free paclitaxel. Method AuNPs were synthesised by sodium citrate reduction method described by Grabar et al. (Grabar, 1995). Then, a 16-Mercaptohexadecanoic acid (16-MHDA) solution was added to the mixture to functionalise the AuNPs surface. After that, a paclitaxel solution was added and conjugated to the functionalised AuNPs using N-(3a) 1705

1637

2849

2920

b)

Transmittance (%)

IS03 Targeted alpha therapy for metastatic melanoma

1402

3551-3097

1781 2849 2920 3551-3323

1081 1637

1230

1705

3216-3097

c)

1402

950 1250

2878

2920

1717 1404

1100

1637 3557- 3348

4500 4000 3500 3000 2500 2000 1500 1000 500

Wavelength(cm-1) Fig. 1 FTIR spectra of (a) 16-MHDA @ AuNPs, (b) NHS-16-MHDA @ AuNPs and (c) 16-MHDA@ Au NPs- paclitaxel conjugates

Relave Survival (%)

Australas Phys Eng Sci Med

Response of T47D cell line to 24h treatment with conjugate B 150 100 50 0 0

0.0068

0.013

0.027

0.054

0.1085

Concentraon of Conjugate B (nM)

Fig. 2 Response of T47D cells to 24 h treatment with 16-MHDA@ AuNPs-paclitaxel conjugates Dimethylaminopropyl)- N0 -ethylcarbodiimide hydrochloride (EDC), and N-Hydroxysuccinimide (NHS) coupling reaction. Results Figure 1 shows the FT-IR spectra of 16-MHDA functionalised AuNPs capped paclitaxel conjugates (16-MHDA@AuNPspaclitaxel). These results confirm the successful conjugation between paclitaxel and the functionalised AuNPs. A cytotoxicity assay for 16-MHDA@AuNPs-paclitaxel conjugates was performed (Fig. 2). T47D cells were treated with five different concentrations of 16-MHDA@ gold nanoparticles-paclitaxel conjugates for 24 h. T47D cell number was reduced significantly compared to the untreated control. Conclusion The 16-MHDA@AuNPs-paclitaxel conjugates were synthesised as drug delivery system for breast cancer treatment. The results revealed T47D cells were killed more effectively after treatment with this system compared with paclitaxel alone.

of Physics, University of Sydney, NSW 2050, Australia. 3GBM Department, Polytech Marseille, France Introduction Gold nanoparticles [GNP] have been used to enhance the local dose delivered in brachytherapy [Ngwa 2012], however consequential biological enhancement is not well understood. Although the manner in which the dose enhancement depends on beam energy and nanoparticle size has been modelled using MonteCarlo simulations, the effects of particle distribution has not been systematically studied. This is an important factor because different studies report a range of responses, which are not always consistent with Monte-Carlo predictions. The aim of this study is to use scintillation dosimetry to show how the distribution of the nanoparticles can affect dose enhancement. Method An organic scintillator composite, incorporating GNPs was made using commercially available scintillator BC498. The light emitted during irradiation of the scintillators was used to evaluate enhancement. The samples were exposed to radiation beams with a range of energies (50–280 kV). The scintillation light was coupled into a PMMA fibre optic and transferred to a photo multiplier tube (PMT) for readout. The PMT was calibrated to dose after subtracting the background signal. Readings were corrected for light absorbed by the nanoparticles. A theory was developed using photon absorption cross sections (Berger et al. 1998).

References 1. Li, N. et al. (2014) Polysaccharide-gold nanocluster supramolecular conjugates as a versatile platform for the targeted delivery of anticancer drugs. 2. Syed, A. et al. (2013) Extracellular biosynthesis of monodispersed gold nanoparticles, their characterization, cytotoxicity assay, biodistribution and conjugation with the anticancer drug Doxorubicin, Nanomedicine & Nanotechnology. 3. Wang, X. et al. (2008) Application of nanotechnology in cancer therapy and imaging. 4. Ochekpe, N. et al. (2009) Nanotechnology and drug delivery part 1: background and applications. 5. Lee, J. et al. (2014) Gold nanoparticles in breast cancer treatment: promise and potential pitfalls. 6. Kumar, A. et al. (2012) Gold nanoparticles functionalized with therapeutic and targeted peptides for cancer treatment. 7. Pissuwan, D. et al. (2011) Review: the forthcoming applications of gold nanoparticles in drug and gene delivery systems. 8. Murawala, P. et al. (2014) In situ synthesized BSA capped gold nanoparticles: Effective carrier of anticancer drug Methotrexate to MCF-7 breast cancer cells. 9. Grabar, K.C. et al. (1995) Preparation and characterization of Au colloid monolayers.

O010 Scintillation dosimetry of dose enhancement with nano particles Natalka Suchowerska1, Laura Toussaint1,2,3, David R. McKenzie2, Paul Z. Y. Liu2 1 Department of Radiation Oncology, Chris O’Brien LifeHouse, NSW 2050, Australia. ([email protected]). 2School

Results The response of BC498/Gy as a function of beam energy for different volume fractions of GNPs, increased with decrease in energy, reaching a maximum and then decreased with further reduction in beam energy. This is unexpected from Monte-Carlo simulation, but can be explained using mathematical modelling of dose reabsorptions of neighbouring particles. Conclusion We have made a direct measurement of the dose enhancement caused by gold nanoparticles in a medium under irradiation. The dose was enhanced preferentially at lower energies, but not as predicted by a simple theory that treats each particle as an independent scatterer. This conclusion is independent of the properties of living cells. References 1. WNgwa et al., Gold nanoparticle-aided brachytherapy with vascular dose painting: Estimation of dose enhancement to the tumor endothelial cell nucleus, Medical Physics 39, 392 (2012) 2. M. J. Berger et al.,The National Institute of Standards and Technology (NIST), XCOM: Photon Cross Sections Database, http://www.nist.gov/pml/data/xcom/(1998)

123

Australas Phys Eng Sci Med

O011 Development of a transmitted alpha particle microdosimetry using A549 cells and Ra-223 source for targeted alpha therapy

O012 Assessment of dose enhancement caused by goldnanoparticles irradiated with protons, as measured with alanine/EPR dosimetry

R. A. L. Darwish1,2, E. Bezak2,3, A. H. Staudacher4,5

C. L. Smith1, T. Takahiro2, M. Geso1

1 Department of Medical Physics, Royal Adelaide Hospital, Adelaide, Australia. 2School of Physical Sciences, University of Adelaide, Adelaide, Australia. ([email protected]), ([email protected]). 3School of Health Sciences, University of South Australia, Adelaide, Australia. 4Translational Oncology Laboratory, Centre for Cancer Biology, SA Pathology, Adelaide, Australia. 5School of Medicine, University of Adelaide ([email protected])

1 School of Medical Sciences, RMIT University, Australia. ([email protected]), ([email protected]). 2Hyogo Ion Beam Medical Centre, Japan. ([email protected])

Introduction Novel microdosimetry techniques are required to estimate a-particle absorbed dose in targeted alpha therapy (TAT) (AL Darwish et al., 2015). In this work, a design of transmitted a-particle microdosimetry system based on semiconductor pixelated radiation detector Timepix (Amsterdam Scientific Instruments) is presented and assessed using A549 lung carcinoma cells and Ra-223 source. Methods A549 cells were seeded at 15,000 cells in transwell inserts (0.14 cm2). The transwell system used, consisted of two compartments: a larger well containing an evaporated Ra-223 source and an insert with a 10 lm thick polycarbonate membrane to which the cells adhered. The transwell system was positioned in front of a Timepix detector and transmitted a-particles were recorded. The quantitative correlation between the distribution of a-particle hits and cell damage was investigated for 0–3 h irradiation times. To assess the cell radiation damage, unirradiated (controls) and irradiated cell monoloayers were stained for the DNA double-strand break (DSB) maker c-H2AX and cell nuclei were counterstained with DAPI. The number of DNA DSB per cell after irradiation and the total damaged cell count were determined. The absorbed dose was estimated using SRIM-2008 Monte Carlo code, and its relationship with DNA DSB and the recorded a-particle hits was investigated. Results c-H2AX staining showed that the incidence of C 6 DSBs in an irradiated cell was *3 times higher after 3 h irradiation compared to 1 h irradiation as shown in Fig. 1a. These numbers correspond to more than 40,000 and 14,000 absorbed a-particles in the irradiated cells. The calculated absorbed dose to cells and medium was 1.7 and 5.1 Gy for 1 and 3 h irradiations respectively. The relationship between the absorbed dose and the DNA DSB cell damage is presented in Fig. 1b. Conclusion Timepix can be used for transmitted microdosimetry, providing good resolution of detected a-particles.

Fig. 1 c-H2AX foci number per 100 cells after 1 and 3 h irradiation. b Absorbed dose to cells and the medium and DNA DSB cell damage relationship References 1. AL Darwish, R., et al., Autoradiography imaging in targeted alpha therapy with Timepix detector. Comput Math Methods Med, 2015. 2015: p. 612580.

123

Introduction Due to discrepancies between dose enhancement effects of gold-nanoparticles (AuNPs) irradiated with protons in animal studies [1], that report *20 % and Monte Carlo simulations [2], [3] which reported minimal dose enhancement, we used alanine doped with AuNPs to determine the dose enhancement caused when irradiated with various beam types (including protons) over a therapeutic dose range. Method Alanine was impregnated with 5 nm AuNPs (3 % w/w) and added within a wax pellet (dimensions 10 9 5 9 5 mm) and placed with control pellets, containing alanine, and exposed to various radiations: low energy (kV ranges) and high energy (MV) X-rays, electrons, and protons. Nominal doses received ranged from 2–20 Gy (within clinical range), and the Electron Paramagnetic Resonance (EPR) spectra of the irradiated samples was recorded. Results Dose enhancement by AuNPs for kV X-rays was *60 %, with smaller dose enhancements observed for 6 MVX X-rays (* 10 %). Dose enhancement was minimal for both 6 MeV electrons (*5 %) and \5 % for 150 MeV protons. Our proton results validate the latest Monte Carlo simulations, which are significantly less than that reported in cell and animal studies (* 20 %). We attribute this difference to the fact that alanine only measures the levels of free radicals generated by the inclusion of AuNPs and not the redox type radicals (such as reactive oxygen species) generated from aqueous media in cells. Conclusion Dose enhancement caused by 5 nm AuNPs with radiotherapy type proton beams has been found to be about 5 % as determined when using alanine/wax as both a phantom and dosimeter. This agrees well with the latest Monte Carlo simulation results for similar sized gold-nanoparticles. Furthermore, our results for both low and high energy X-rays are validated against published data for in vitro studies. References 1. Kim, J.K., et al., Enhanced proton treatment in mouse tumors through proton irradiated nanoradiator effects on metallic nanoparticles. Phys Med Biol, 2012. 57(24): p. 8309–23. 2. Jeynes, J.C., et al., Investigation of gold nanoparticle radiosensitization mechanisms using a free radical scavenger and protons of different energies. Phys Med Biol, 2014. 59(21): p. 6431–43. 3. Walzlein, C., et al., Simulations of dose enhancement for heavy atom nanoparticles irradiated by protons. Phys Med Biol, 2014. 59(6): p. 1441–58

O013 Perfluorocarbons and lipid emulsion as novel treatments for Cerebral Radiation Necrosis Lisa Anne Feldman1, Bruce Spiess2, William C. Broaddus1, Wen Wan3, Joel R. Garbow4 1 Department of Neurosurgery, Virginia Commonwealth University, Richmond VA 23223. ([email protected]), ([email protected]). 2Department of Anaesthesia, Virginia Commonwealth University, Richmond VA 23223.

Australas Phys Eng Sci Med ([email protected]). 3Department of Biostatistics, Virginia Commonwealth University, Richmond VA 23223. ([email protected]). 4 Department of Radiology, Washington University in St. Louis, St. Louis MO. ([email protected]) Introduction Cerebral radiation necrosis (RN) is a serious and doselimiting complication of stereotactic radiosurgery (SRS) that results in debilitating symptoms, including seizures, neurological deficits, and death. Current treatments for RN remain suboptimal, due to marginal therapeutic effects or undesirable side effects. As hyperbaric oxygen therapy is a partially effective treatment for RN, we proposed to explore using Perfluorocarbons (PFCs) as a novel approach to deliver oxygen therapy for RN. Methods Studies employed a mouse model of RN developed in our laboratory, in which mice are irradiated with a single-fraction, hemispheric 50-Gy radiation dose using the Leksell Gamma Knife (GK) PerfexionTM (Elekta; Stockholm, Sweden) (1). Mice received one of three intravenous treatments (n = 8/cohort) at the first sign of radiation injury (4 weeks post irradiation), as appreciated by contrastenhanced T2-weighted MRI: high dose of 60 % PFC (5.4 g/kg), low dose of 60 % PFC (1.8 g/kg), or 20 % lipid emulsion (Intralipid) control (9 ml/kg). Animals received 6 treatments, 3 days apart, and were housed in a hyperoxic environment (50 % fraction inspired oxygen (FiO2)) throughout treatment. Gadolinium-enhanced, T2weighted MRI scans were obtained at weeks 3, 4, 5, 6 and 7 following irradiation. RN lesion size was analysed by MATLAB-based software that segments hyperintensity in MR images. Lesion sizes were analysed using repeated measure ANOVA. Results There were no statistical differences in RN lesion size between groups at weeks 3,4,or 5 following GK SRS, however all experimental groups showed less RN than historical controls at week 6 (all p-values less than 0.0001). At week 7, RN lesions are significantly smaller in low-dose PFC (p = 0.0270) and Intralipid (p = 0.0038) groups, compared to historical controls. Conclusion IV administration of low-dose PFC or lipid emulsion in a hyperoxic environment appears efficacious and may be a novel treatment for mitigating the development of RN.

RN Lesion Volume Size Lesion (mm^3)

50 40 30

Low

20

High Intralipid

10 Control

Introduction At Nepean Cancer Care Centre, Frameless Sterotactic Radiosurgery (FSRS) was commissioned for treatment of patients with multiple metastatic tumours, using a single isocentre. Noncoplanar VMAT beams were planned on the Philips Pinnacle3 TRS. All patients were treated on an Elekta Synergy fitted with an Agility MLC and with XVI/iView imaging. The main aim of commissioning was to assess the dosimetric and geometric accuracy of the linear accelerator as used in this technique. Method Dosimetry involved testing of output factors, PDD and profile assessment for small treatment fields. Pinnacle3 MLC round end offset table was re-examined. Isosphere for gantry, collimator and couch rotation was tested. Couch isocentre and rotational position accuracy was examined. Patient plans were measured on ScandiDos Delta4 with couch zero due to setup limitations with this device. A point dose measurement was made using an Extradin A16 micro ion chamber in a cylindrical water phantom at the planned treatment couch angle. Results Small field dosimetry assessment indicates 1.0 % agreement between planned and measured doses for a 2 9 2 cm field. The Pinnacle3 MLC round end offset table was refined to further reduce the discrepancy between planned and measured dose. The iso-sphere for gantry and collimator are 1.0 mm, while the couch isosphere was found to be 2.0 mm over the full range of couch rotation. To reduce the effect of the couch rotational error in a clinical setting we limited the allowable couch angles available for planning to ± 45 degrees. Couch rotational accuracy was found to be ± 0.1 degree. Delta 4 QA shows gamma confidence range from 94.7 % to 100.0 % in 1 mm/3 % with 10 % threshold on five patient plans after adjustment of the MLC round end offset table. All point dose measurements made with the A16 micro chamber were within 2.0 % of planned doses. Conclusion It is possible to implement and treat non-coplanar treatment of FSRS, with dosimteric and geometric accuracy on an Elekta Synergy machine, with refinements and restrictions. References 1. Timothy D. Solberg, et al., ‘‘Quality and safety considerations in stereotactic radiosurgery and stereotactic body radiation therapy’’ Practical Radiation Oncology (2011) 2. Fraass B, Doppke K, Hunt M, et al., ‘‘American Association of Physicists in Medicine Radiation Therapy Committee Task Group 53. Quality assurance for clinical radiotherapy treatment planning’’ Med Phys. 1998;25(10):1773–1829. 3. Ezzell, Garry A., et al., ‘‘IMRT commissioning: Multiple institution planning and dosimetry comparisons, a report from AAPM Task Group 11900 Med Phys. 2009:36(11):5359–5373

0 Week 3

Week 4

Week 5

Week 6

Week 7

Time

References 1. X. Jiang, et al. Anti-vegf antibodies mitigate the development of radiation necrosis in mouse brain. Clinical Cancer Research, in press (2014), Epub 2014/03/22.

O015 Use of linac based stereotactic radiotherapy for intra ocular cancers—what treatment margins should we apply? Rachitha Antony1, Alan Herschtal1, Stephen Todd1, Claire Phillips1, Annette Haworth1 1

O014 Commissioning a non-coplanar VMAT technique for use in frameless brain stereotactic radiosurgery on an elekta linear accelerator J. Luo, S. White Nepean Cancer Care Centre, NSW, Australia. ([email protected]), ([email protected])

Peter MacCallum Cancer Centre Melbourne Victoria. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]) Introduction The Eye Tracker system, for stabilising and monitoring eye movements for linac-based stereotactic radiotherapy associated with the mobile eye was developed based on the design of Petersch (1). We report our preliminary estimates of the margin required to

123

Australas Phys Eng Sci Med

Dimension

Margin

Lateral

0.41, [0.23, 0.91]

Superior-inferior

0.5, [0.28, 1.22]

account for inter- and intra-fraction eye motion based on data from 7 consecutive patients. Methods Patients were immobilised in a head and neck mask and were required to fixate on a light source. A camera system monitored eye movement relative to CT simulation baseline measurements. Treatments were delivered in 5 or 30 fractions. The Exactrac system (Brainlab, Feldkirchen, Germany) combined with the Varian TrueBeamSTx (Varian Medical Systems, Palo Alto, CA) confirmed standard pre- and intra-treatment setup. Displacement/rotation of the image of the pupil/iris was determined in the lateral and superiorinferior directions using the video display. The equation of van Herk (2) was applied to estimate the margin required to account for interand intra-fraction eye movement. Results The average displacement in both directions was 0.1–0.2 mm (0.36 mm SD). All patients maintained a position within 1 mm of the intended position. Based on a Bayesian estimation of the systematic and treatment errors, accounting for displacements in 2-planes and a penumbral width (distance between 80–90 % isodoses) of 1.3 mm, the calculated margins and credible intervals to achieve coverage of the GTV with the 95 % isodose in 90 % of patients is shown in Table 1. Conclusion Small random and systematic uncertainties due to interand intra-fraction movement of the eye were achieved with the Eye Tracker. Estimated margins are small (\1 mm) but need to be considered in addition to contouring and treatment delivery uncertainties. We acknowledge the funding support of The Peter MacCallum Foundation.

Method The response on the central axis was measured using commercial diodes. Jaw positions were varied from 0.5 mm 9 0.5 mm to 40 cm 9 40 cm and were kept square for all measurements. The variation in the measured output was recorded for each cone. Results The figure below shows the result for the 5 mm cone. Smaller than a 1 cm 9 1 cm field results in the signal dropping sharply due to the jaws stopping the beam from passing through the cone. Higher than a 10 cm 9 10 cm field results in the signal increasing due to leakage around the edges of the cone. Based on these measurements a jaw setting of 2 cm 9 2 cm was chosen for the 5 mm cone. Conclusion Optimum jaw positions need to be determined when using stereotactic cones if they are not interlocked and automatically positioned by the linear accelerator. Square jaw settings have been determined for the suite of cones purchased by ARPANSA so as to minimize their influence when using each cone. 5mm cone P diode signal (nC)

Table 1 Expected van Herk margin (mm) and 95 % credible intervals

0

2

4

6

8 10 12 14 16 X and Y jaw position (cm)

18

20

22

O017 Commissioning a dedicated planning system for the treatment of brain metastases: early experience with BrainLab Elements

References

J. Kenny1, D. Jolly1, S. Atkins1, E. Jhala1, S. Towns1, M. Dally1

1. Petersch B, Bogner J, Dieckmann K, Potter R, Georg D. Automatic real-time surveillance of eye position and gating for stereotactic radiotherapy of uveal melanoma. Med Phys. 2004 Dec;31(12):3521–7. 2. van Herk, M., et al. (2000). The probability of correct target dosage: dose-population histograms for deriving treatment margins in radiotherapy. Int J Radiat Oncol Biol Phys 47(4): 1121–35.

Epworth Radiation Oncology, Epworth Healthcare, Melbourne, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected])

O016 The effect of jaw positions on stereotactic cone output C. P. Oliver1, D. J. Butler1 1 Radiotherapy section, ARPANSA, Australia., ([email protected]), ([email protected])

Introduction The ARPANSA Elekta Synergy linear accelerator was recently modified to allow the attachment of Elekta stereotactic cones to the treatment head. Five circular cones were procured with diameters 50, 15, 10, 7.5 and 5 mm. The presence of the cones is not sensed by the accelerator and any jaw position can be set with any cone. Jaw positions needed to be determined for precise standards measurements using the cones at ARPANSA.

123

Introduction Linac based stereotactic radiosurgery has been used to treat isolated brain metastases, however has had limited use for multiple lesions as an alternative to whole brain radiation therapy. The BrainLab Elements (BrainLab AG, Munich) Multi Metastases Element (MME) utilises what could be considered a multi-target dynamic conformal arc (DCA) delivery. MME enables the delivery of stereotactic doses to multiple brain metastases simultaneously, with high conformity and efficient delivery. Method Elements with MME v1.0.0.69 was installed on an existing iPlanNet server and configured for a Varian Novalis Tx 6MV beam. Dose calculations use the existing beam profile(s) for the Pencil Beam Convolution algorithm within iPlan RT Dose v4.5.3, for conventional (non-FFF) beams only. Due to lack of a physics quality assurance application, MME deliveries were simulated in iPlanRT Dose using an IMT Head Phantom (Integrated Medical Technologies, Troy NY) and spherical targets of 5–30 mm diameter. DCA fields with both central and 60 mm off-axis isocentre were used.

Australas Phys Eng Sci Med Targets of B 10 mm diameter were measured with film using a small region of interest centred on maximum dose. The 20 and 30 mm diameter targets were measured with a centrally placed IBA CC13 ionisation chamber. All plans were independently re-calculated using Mobius3D (Mobius Medical Systems, Houston). Results The agreement between film measured and planned dose was in the range -0.4 to +3.8 %. 2D gamma tests with a 3 %/1.0 mm criteria returned an average 90 % pass rate (range 78.3–100 %). The CC13 measured and planned dose agreement ranged from -1.6 to +0.1 %. Mobius 3D agreement ranged from -8.8 to +0.5 % and showed a strong tendency toward poor agreement with decreasing target volume. Conclusion The BrainLab Multi Metastases Element has been commissioned and clinically implemented at Epworth Radiation Oncology. Treatment plan dosimetry has been verified with measurement and independent dose calculation.

O018 Development of an end-to-end audit process for stereotactic radiosurgery J. Morales1,2, M. Butson1, S. Alzaidi1, M. O’Connor, G. Perez1, R. Hill1 1 Department of Radiation Oncology, Chris O’Brien Lifehouse, Sydney, Australia. 2School of Chemistry, Physics and Mechanical Engineering, Queensland University of Technology, Queensland, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]), (gino,[email protected]), ([email protected])

Introduction Stereotactic radiosurgery (SRS) requires high accuracy and precision due to the large radiation doses and small tumour volumes for these patients. An assessment of the overall accuracy for a patient can be determined by end-to-end testing using appropriate phantoms. The purpose of this work is to develop a suitable end-toend test for auditing SRS treatments. Method All SRS treatments in this department are delivered on a Novalis Trilogy linear accelerator using MLCs and cones along with ExacTrac imaging systems for patient setup. End-to-end testing was performed using a Lucy SRS phantom and CIRS anthropomorphic phantoms. IR markers were positioned onto the phantoms to allow for setup with ExacTrac. CT scans were taken of the phantoms using a 1 mm slice thickness and imported into the BrainLab iPlan planning system. Patient plans were transferred onto the phantoms for dose calculations. Gafchromic EBT3 film sheets were placed in cassettes (with fiduicial markers) and positioned inside the phantoms. Treatment plans were transferred to the ExacTrac system for automatic setup using the IR markers and DRR images and positional accuracy for the IR markers of 0.5–1 mm. The treatment plan was delivered with the 6 MV SRS beam and the subsequent dose distributions measured using an Epson desktop scanner. Results For the treatment with the 4 mm diameter BrainLab cone, the shifts between the planning and measured dose profiles were 0.4 mm in the Left/Right direction and 0.2 mm in the Superior/Inferior direction. These are well within the tolerances for ExacTrac based on the manufacturer’s specifications. Conclusion We have developed a suitable end-to-end audit for testing of SRS treatments which is available for clinical use. Further testing will include the impact of image fusion with multiple image modalities in the planning process.

O019 The Australian national collaborative academic Medical Physics Network: structure, activity and outcomes David Thwaites1, Martin Carolan2, Andrew Fielding3, Rick Franich4, Mike House5, Joerg Lehmann1, Peter Metcalfe6, Scott Penfold7, Anatoly Rosenfeld6 1

University of Sydney. ([email protected]). Wollongong Hospital. 3Queensland University of Technology, Brisbane. 4Royal Melbourne Institute of Technology, Melbourne. 5 University of Western Australia, Perth. 6University of Wollongong. 7 University of Adelaide 2

Introduction A national Australian inter-university medical physics (MP) group was formed in 2011/12, supported by Department of Health Better Access to Radiation Oncology (BARO) seed funding. Core membership includes the six universities providing postgraduate MP courses. Objectives include increasing capacity, development and efficiency of national academic MP structures/systems and hence supporting education, clinical training and research, for the MP workforce support. Although the BARO scheme focuses on Radiation Oncology, the group has wider MP interests. Methods Two further BARO seed grants were achieved: 1) for networked academic activities, including shared-resource teaching, e.g. using virtual reality systems; MP outreach to schools and undergraduates; developing web-based student and registrar education/ resources, etc.; and 2) for conjoint ‘translational research’ posts between universities and partner hospitals, to clinically progress advanced RT technologies and to support students and registrars. Each university received 0.5 FTE post from each grant over 2 years (total: $1.75 M) and leveraged local additional partner funds. Results Total funding: $4–5 M. Overall there have been approximately 35 (mainly overseas) postholders bringing specific expertise, beginning in early 2013. Periods in Australia have been from 0.25–2 years (median = 0.5–1). As well as the education activities, research projects include lung/spine SBRT, 4D RT, FFF beams, technology assessment, complex treatment planning, imaging for radiation oncology, DIR, adaptive breast, datamining, radiomics,etc. Observed positive impacts include: increased interest in MP courses, training support, translational research infrastructure and/or clinical practice in the hospitals involved, plus increased collaboration and effectiveness between the universities. Posts are continuing beyond grant end using leveraged funds, providing the basis for sustainability of some posts. Conclusion The BARO-funded projects have cost-effectively produced a range of positive impacts on training, research and practice in hospitals and between universities. The evaluation of the specific post roles and activities, and their outcomes, has produced focused recommendations on continuation and sustainability.

O020 Initiatives of the ACPSEM Queensland branch S. B. Crowe1, P. Charles2, T. Ireland3, B. Sutherland4, N. Middlebrook4, K. Biggerstaff2, M. Irvine3, S. Critchley5, T. Markwell6, T. Kairn4 1 Royal Brisbane & Women’s Hospital, Brisbane, Australia. ([email protected]). 2Princess Alexandra Hospital, Brisbane, Australia. ([email protected]),

123

Australas Phys Eng Sci Med ([email protected]). 3Biomedical Technology Services, Gold Coast, Australia ([email protected]), ([email protected]). 4Genesis Cancer Care Queensland, Gold Coast, Australia. ([email protected]), ([email protected]), ([email protected]). 5Radiation Health, Department of Health, Brisbane, Australia ([email protected]). 6Mater Cancer Care Centre, Brisbane, Australia. ([email protected]) Introduction The ACPSEM Queensland Branch (ACPSEMQ) aims to provide opportunities for local members that complement the broader benefits of ACPSEM membership. Our activities include the branch symposium (with invited speakers), a half-day ‘‘Winter School’’, three ‘‘Progress and Research in Medical Physics’’ meetings annually; and the ‘‘Fitch Award’’—all promoted via the branch website (acpsemq.org.au) and newsletter. Some of these events are the result of a policy to introduce at least 1 major initiative every year. Method The branch committee (a group of no more than 10 representatives) meets quarterly to discuss activities and propose initiatives. As the group that runs ACPSEMQ events, we organise sponsorship, venue hire, travel arrangements and accommodation. Where new initiatives are adopted, the committee has, both formally and informally, obtained feedback from members. Attendance at events is recorded for discussion. Photos and a recap are typically distributed afterwards for those members unable to attend. Results The ACPSEMQ members have responded favourably to the introduction of initiatives, with attendance at ACPSEMQ events improving (RSVPs increasing 76 % between March 2013 and February 2015). Attendees at events have indicated that they found the presentations and discussions both informative and fruitful. This has also provided increased opportunities for networking and knowledge sharing. Conclusion We believe the concept of introducing an additional initiative each year helps keep engagement with the members dynamic and interesting; and thus far feedback from the membership has been positive. Acknowledgements This work was supported by past members of the ACPSEM Queensland Branch committee and the wider membership.

TrueBeam 6MV 28 cm2 parallel opposed photon beams were arranged with the isocentre in the geometric centre of the cage. Mouse phantom dose distributions were calculated to give 2.0 Gy to the centre of one of the mouse phantoms with Eclipse AAA and Acuros and compared to VIMC with calculations performed using identical monitor units as the respective TPS plans. Results Mouse Phantom

AAA Mean(Range) Dose/Gy

AAA Plan VIMC Mean(Range) Dose/Gy

Acuros Mean(Range) Dose/Gy

Acuros Plan VIMC Mean(Range) Dose/Gy

1

2.00(1.84–2.06)

2.03(1.92–2.08)

1.99(1.87–2.03)

1.94(1.82–1.99)

2

1.99(1.85–2.07)

2.03(1.94–2.07)

1.99(1.90–2.05)

1.94(1.85–1.99)

3

1.99(1.84–2.06)

2.03(1.93–2.07)

1.99(1.90–2.03)

1.94(1.85–1.98)

VIMC and Acuros doses are dose to water values

Conclusion It appears that Eclipse TPS calculations are accurate enough to perform mouse TBI calculations for the geometry in this work. This makes it feasible for general use in centres where no sophisticated TPS dose calculation algorithms (e.g. Acuros of Monte Carlo) are available. References 1. Zavgorodni S, Bush K, Locke C and Beckham W 2007 Vancouver Island Monte Carlo (VIMC) system for radiotherapy treatment planning dosimetry and research. Radiotherapy & Oncology 84 Supplement 1, S49

O022 Dosimetric assessment for whole brain irradiation in mice using a gamma irradiator N. Schleich1,2, P. Herst1 1

Department of Radiation Therapy, University of Otago Wellington, New Zealand. 2Blood & Cancer Centre, Wellington Hospital, Wellington, New Zealand. ([email protected]), ([email protected])

O021 Mouse total body irradiation dosimetry W. A. Beckham1,2, L. DeVorkin1 1 British Columbia Cancer Agency – Vancouver Island Centre. ([email protected]), ([email protected]) 2 University of Victoria Physics & Astronomy Department

Introduction Laboratory mouse total body irradiation (TBI) has improved our ability to study a number of radiation-induced responses in a physiological setting. In addition to research examining the effects of TBI on acute radiation and radioadaptive responses, it is also used in the fields of cancer biology, immunology and hematopoietic cell studies. This work describes a general set-up and precise dosimetry for mouse TBI using a commercial treatment planning system (TPS) compared with Monte Carlo methods. Such an approach was considered necessary due to the potentially challenging geometry of the set-up. Method A 21.5 cm diameter commercial mouse ‘‘pie cage’’ was sandwiched between two 30 9 30 cm2 9 5 cm thick slabs of solid water. 3 acrylic cylindrical mouse phantoms were placed in the pie cage and CT scanned. Data was uploaded to the Varian Eclipse TPS and the Vancouver Island Monte Carlo (VIMC) system1. Varian

123

Introduction Glioblastoma multiforme is an aggressive brain cancer which is highly resistant to ionising radiation (Stupp et al. 2009; McCord et al. 2009). Our recent research has investigated the radiosensitizing effect of high dose ascorbate in an intracranial mouse model (Herst et al. 2012; Grasso et al. 2014). This presentation discusses mouse body shielding and dose assessment for whole brain irradiation in live mice using a gamma irradiator. Method A protocol was developed for irradiation of live mice using the Gammacell 3000 Elan irradiator (Best Theratronics) designed for irradiating blood products using a 48 TBq 137Cs source. A custombuilt mouse holder with lead shielding was designed for whole brain irradiation, whilst protecting the mouse body and maintaining body temperature. Radiochromic film and thermoluminescent dosimeters (TLDs) were used to assess the dose distribution and to measure the absorbed dose delivered to the head and different parts of the body. Calibrations of the TLDs were performed using a 6 MV photon beam, and a 137Cs source (Kron 1994, Haworth et al. 2013). Results The results allowed quantitative assessment of the dose distribution over the mouse including scatter inside the shielding, and validated the shielding design. Analysis of the results, including an estimate of the uncertainties, indicates an absorbed dose of 4.5 ± 0.3 Gy delivered to the brain region and a mean of

Australas Phys Eng Sci Med 0.6 ± 0.04 Gy along the mouse body within the shielding. The difference between measured and reported doses was quantified. Conclusion Dosimetry for partial body irradiation utilising the gamma irradiator is challenging due to the size and location of the target volume and the fact that the system is designed to produce a near uniform dose profile. Gafchromic film and TLD measurements confirmed the functionality of the mouse holder, the shielding design and the irradiation protocol. References 1. Stupp R, Hegi ME, Mason WP, et al. Lancet Oncol. 2009;10:459–66. 2. McCord AM, Jamal M, Williams ES, et al. Clin Cancer Res. 2009;15:5145–53. 3. Herst PM, Broadley KWR, Harper JL, McConnell MJ. Free Radical Biol Med. 2012;52(8):1486–93. 4. Grasso C, Fabre MS, Collis SV, Castro ML, Field CS, Schleich N, et al. Front Oncol. 2014;4:356. doi:10.3389/fonc.2014.00356. 5. Kron T. Australas Phys Eng Sci Med. 1994;17(4):175–99. 6. Haworth A, Butler DJ, Wilfert L, Ebert MA, Todd SP, Hayton AJM, Kron T. J Appl Clin Med Phys. 2013;14(1):258–72.

O023 Development of a quality control program for spectral micro CT imaging

windows 20–140 keV. QC results were used to establish suitable scan settings, data reconstruction parameters and analysis approaches, to acquire baselines and to perform constancy checks. Results obtained with the customized phantoms have assisted in rapid diagnosis of non-ideal behaviour, leading to improved image quality. Pre-clinical research examples of scans performed before and after introduction of the QC program demonstrate the necessity of checking each link in the imaging chain. Conclusion Imaging QC necessitates robust procedures that are reproducible whilst being adequately sensitive to changes in the parameter to be tested. Phantoms and procedures suitable for micro spectral CT were developed and tested for this purpose. References 1. Anderson NG, Butler AP, Scott N, et al. Spectroscopic (multienergy) CT distinguishes iodine and barium contrast material in mice. Eur Radiol. 2010;20:2126–34. 2. Aamir R, Chernoglazov A, Bateman CJ, et al. MARS spectral molecular imaging of lamb tissue: data collection and image analysis. JINST. 2014;9: P02005, 1–10.

O024 Optimization of Pinnacle parameters for improving QA results in VMAT R. C. Joshi, Julia Green, Sarah Wong

N. Schleich1, R. Aamir2, S. M. Midgley3, S. Bell4, N. Anderson2, A. Butler2,5,6,7 1 Department of Radiation Therapy, University of Otago Wellington, New Zealand. ([email protected]). 2Department of Radiology, University of Otago, Christchurch, New Zealand. ([email protected]), ([email protected]). 3School of Physics, Monash University, Melbourne, Australia. ([email protected]). 4MARS Bioimaging, Christchurch, New Zealand. ([email protected]). 5Department of Electrical and Computer Engineering, University of Canterbury, Christchurch, New Zealand. 6MARS Bioimaging, Christchurch, New Zealand. 7European Organisation for Nuclear Research, Geneva, Switzerland. ([email protected])

Introduction Our NZ based research team is developing the MARS spectral scanner, which is a small-bore spectral computed tomography (CT) system for pre-clinical use (Anderson et al. 2010; Aamir et al. 2014). The MARS scanners feature a rotating microfocal X-ray source and semiconductor detector array with photon counting and energy resolving capabilities. The field of view is approximately 100 mm diameter by 300 mm long, achieving 50–100 lm spatial resolution at the rotation axis depending on the detection system used. Quality control (QC) testing is a vital link in the imaging chain, monitoring and quantifying the performance of each subsystem. Micro CT requires specifically designed test objects and protocols for this task. The energy resolving capabilities of individual pixels requires careful balancing across the detector array and global testing. Method The methodologies for testing medical CT were adapted, requiring the design of test objects appropriate to micro CT. Protocols and algorithms were developed for rapid measurement of image noise and uniformity, high contrast spatial resolution, low contrast detail threshold and other parameters. Results MARS spectral CT scans of the phantoms underwent 3D algebraic reconstruction delivering volumetric data in five energy

Senior Medical Physicist, AWCCC, Northern Territory Radiation Oncology, NT 0810, Australia. ([email protected]), ([email protected]), ([email protected]) Introduction Considering the complexity of Volumetric modulated arc therapy (VMAT) delivery, optimization of linac and treatment planning system (TPS) parameters is essential for achieving desired quality assurance (QA) results. Accurate delivery of planned dose may be affected by parameters including variable dose rate, gantry speed, leaf constraint, minimum leaf segments, minimum segment area and couch modelling. This work demonstrates the effect of leaf speed and couch model on VMAT planned and delivered dose. Method VMAT plans were calculated in Pinnacle, delivered on an Elekta Synergy Linac and QA performed with MapCheck. The effect of leaf constraint on DVH statistics and delivery accuracy was investigated for 0.12, 0.46, and 0.80 cm/deg for a complex head and neck single arc VMAT plan with a 70 Gy prescription. The effect of couch density on QA results was observed for inner couch densities 0–0.15 and outer couch density 0.8 for 5 VMAT and 10 IMRT plans. Results The DVH analysis shows optimal PTV coverage for leaf constraint of 0.46 cm/deg with maximum PTV dose of 73 Gy and mean dose of 70 Gy compared to a maximum 80 Gy and mean 70 Gy and maximum 71 Gy and mean 67.12 Gy for leaf constraint of 0.12 and 0.80 cm/deg respectively. The gamma pass rate was highest for the plan calculated with leaf constraint of 0.46 cm/deg. The average gamma pass rate (3 %/2 mm) of plans calculated with couch density of 0 and 0.15 were 93.62 % and 95.22 % respectively for VMAT plans and 96.09 % and 96.94 % respectively for IMRT plans. Conclusion Leaf constraint set to 0.46 cm/deg achieves better PTV coverage and QA result for VMAT plans compared to other constraints. Inner couch density 0.15 for various VMAT and IMRT fields shows improved QA results. The magnitude of couch effect is less for VMAT arcs compared to Static IMRT fields.

123

Australas Phys Eng Sci Med

O025 Measuring sliding window output factors using MLC files with 2 control points: The impact on the dosimetric leaf gap and IMRT deliveries J. Hindmarsh1, C. Lee1,2, B. Zwan1,2, E. Seymour1, K. Sloan1, R. David2 1

Central Coast Cancer Centre, CCLHD, Gosford, NSW. ([email protected]). 2School of Mathematical and Physical Sciences, University of Newcastle, Newcastle, NSW, Australia. ([email protected]) Introduction The dosimetric leaf gap (DLG) is required in the Eclipse TPS in order to model the rounded leaf tip of the MLC. The DLG can be determined by measuring sliding window output factors (SWOFs). In this work, we report on the difference between measuring SWOFs using MLC files with both 2 and 50 control points (CPs). We show that using MLC files with only 2 CPs results in significant errors in the measured DLG and quantify how these errors translate into differences between planned and delivered IMRT dose distributions. Method Sliding window (SW) MLC fields were created with various gap widths in Shaper and delivered on a Varian Clinac. For each gap width, two MLC files were created consisting of 2 and 50 CPs. The SWOF was measured for each field and subsequently compared to the output factors predicted by the TPS. The DLG was measured for SW files with both 2 and 50 CPs using the method described by Vial et al. (2006). Patient dose distributions were calculated for several clinical IMRT plans using the two different DLG values. Results Figure (a) shows the percentage difference between the measured and TPS-predicted SWOF values with both 2 and 50 CPs. The differences can be attributed to the fact that the leaf position offset table is only applied at each CP. The DLG measured using 50 CPs and 2 CPs is given in Figure (a). Figure (a) displays an example of the dose discrepancy resulting from the use of an incorrect DLG measured using only 2 CPs for an anal canal treatment. Conclusion Measuring SWOFs using MLC files with only 2 CPs results in an incorrect determination of the DLG which can produce clinically significant dosimetric errors in IMRT deliveries. (b)

to minimise the impact of calculation and delivery errors. With QA results being well within our clinical level of acceptability, we would now like to increase plan complexity to improve our organ at risk sparing. Recent measurements have indicated that the settings of the dynamic leaf gap (DLG), the leaf transmission (LT) and target spot size (TSSx and TSSy) in our Eclipse Treatment Planning System (Varian Medical Systems) could be further optimised for the 6 MV beams on our Varian TrueBeam and iX linacs. Their impact on dose accuracy is a convolution of the individual effects. As the net impact is likely to increase with plan complexity we decided to undertake a study to investigate various methods to distinguish and optimise settings for these parameters. Methods The DLG was determined using: 1) dynamic sweeping gap, 2) IMRT test pattern (Van Esch, 2004) and 3) VMAT sweeping gap deliveries. The LT was determined by ionometry with: 1) static MLCs, 2) dynamic IMRT delivery. Time resolved ionometry (Louwe, 2015) was used to verify all parameters in concert. Results Initial measurements indicate: A) the DLG is optimal between 1.0 and 1.2 mm for the TB and 1.4 and 1.6 for the iX; B) the LT is optimal between 1.4 % to 1.6 % for both systems; C) the TSSx appears optimal at 1.75 mm for the iX (Fig. 1). Conclusion The current beam parameters can be adjusted to further improve the accuracy of VMAT deliveries. Time resolved ionometry seems to be the most sensitive measurement method as it can resolve the different effects. References

DLG (mm)

CPs

6 MV

10 MV

2

1.98

2.11

50

1.47

1.61

References 1. Vial, P., et al. (2006). An experimental investigation into the radiation field offset of a dynamic multileaf collimator. Physics in Medicine and Biology 51(21): 5517–5538.

O026 Optimisation of eclipse beam model for VMAT delivery A. J. Williams1, R. J. W. Louwe1, T. W. S. Satherly1 1

Wellington Blood and Cancer Centre (WBCC), Wellington, NZ ([email protected]), ([email protected]), ([email protected]) Introduction When VMAT treatments were introduced at the WBCC in 2010, the plan complexity was kept as low as reasonably practical

123

Fig. 1 Time resolved ionometry plots. Left measured versus calculated dose [cGy] of the VMAT sweeping gap delivery for different DLG and TSSx. DLG = 2.1 mm is the current iX setting. The primary influence points of the various parameters are indicated. Right deviation per control point [% of the integral dose per delivery]

1. Eclipse V11, Varian Medical Systems, Palo Alto, California, USA. 2. Van Esch A, Depuydt T, Huyskens DP., The use of an aSi-based EPID for routine absolute dosimetric pre-treatment verification of dynamic IMRT fields. Radiother Oncol. 2004 May;71(2):223–34. 3. Louwe RJ, Wendling M, Monshouwer R, Satherley T, Day RA, Greig L., Time-resolved dosimetry using a pinpoint ionization chamber as quality assurance for IMRT and VMAT. Med Phys. 2015 Apr;42(4):1625–39. doi: 10.1118/1.4914395.

O027 Department experience in commissioning the Acuros XB algorithm for the eclipse treatment planning system S. Alzaidi, E. Claridge Mackonis, G. Cho, J. Poder, D. Odgers, M. Whitaker, M. Butson, J. Morales, D. Pope, R. Hill 1

Chris O’Brien Lifehouse, Sydney, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected])

Australas Phys Eng Sci Med Introduction The Acuros XB (AXB) algorithm (Varian Medical Systems) is a photon dose calculation algorithm in the Eclipse treatment planning system. Acuros XB is based on an analytical solution of the linear Boltzmann transport equation where the effects of heterogeneities in patient dose calculations are directly accounted for1. This work presents the results of commissioning the Acuros XB algorithm for a range of treatment techniques. Method The setup of the Acuros XB algorithm involved tuning the parameters (source size, MLC transmission) in the beam modelling section. These were verified by comparisons of dose calculations small and large open X-ray fields. Testing of the Acuros XB algorithm included comparisons to measured doses plus doses calculated with the current Analytical Anisotropic Algorithm (AAA) in Eclipse. The basic verification measurements include open field point doses using a ionisation chamber in a water phantom following the IAEA TRS 430 methodology2. More complex testing involved plans recommended in the IAEA TecDoc 1583 using a CIRS thorax phantom3. Tissue inhomogeneity testing was performed using Gafchromic EBT3 film and ionisation chamber measurements in slabs of different materials as well as point dose measurements using the CIRS thorax phantom for IMRT and VMAT plans. Calculation time comparisons between Acuros XB and AAA were also considered. Results For open X-ray fields, the agreement between Acuros XB and AAA was very good. Both algorithms had similar differences from measured data for smaller field sizes. However Acuros XB was found to be up to 2 % more accurate in heterogeneous media for standard field sizes based on the testing in the different phantoms. The downside is that calculation time using Acuros XB algorithm can be significantly larger. Conclusion The Acuros XB calculation algorithm is superior to the AAA algorithm; however time cost of using this algorithm has to be considered.

inhomogeneities and does not give any dose distribution information. Raystation TPS uses Electron Monte Carlo (EMC) calculation on image based datasets. To assess the EMC beam model, several tests were performed based on TG252 for 6, 9, and 12 MeV for 3 matched Elekta machines at 100 cm and 110 cm SSD. The mechanical jaw parameters were verified against the reference machine. The output factors, PDDs and profiles for open and irregular fields as well as oblique incidence were compared to water tank measurements. The TPS dose for several bolus setups, inhomogeneities and clinical plans were compared to diode measurements. Photon-electron junctions were compared using film dosimetry. Gamma analysis was done using MartriXX 2D array for all profiles and clinical plans then further analyzed with an in-house 1D spreadsheet. Transfer of TPS plans to Mosaiq and RadCalc was successful with acceptable workarounds. All point dose comparisons at dmax were within 5 % tolerance. All 2D gamma analyses measured at dmax were[95 %. Some profiles, particularly for large fields, will require improvement in subsequent model revisions due to penumbral mismatches as found in the 1D gamma analysis. The Electron Monte Carlo module within Raystation v.4.0.3.4 is acceptable for clinical use with minor issues that has been reported to Raysearch for future version enhancements. References 1. Almon S. Shiu, Samuel S. Tung, Carl E. Nyerick, Timothy G. Ochran, Victor A. Otte, Arthur L. Boyer and Kenneth R. Hogstrom. Comprehensive analysis of electron beam central axis dose for a radiotherapy linear accelerator. Med.Phys. 21,559 (1994) 2. Gerbi et al., 2009. Recommendations for clinical electron beam dosimetry: Supplement to the recommendations of Task Group 25. Medical Physics, 36, 3239–3279 (2009)

References 1 Failla G A, Wareing T, Archambault Y, Thompson S, ‘‘Acuros XB advanced dose calculation for the Eclipse treatment planning system’’ Palo Alto, CA: Varian Medical Systems; 2010. 2 IAEA Technical Reports Series no. 430, Commissioning and Quality Assurance of Computerized Planning System for Radiation Treatment of Cancer. IAEA, 2004. 3 IAEA-TECDOC-1583, Commissioning of radiotherapy treatment planning system: Testing for typical external beam treatment techniques, Report of coordinated research project (CRP) on development of procedures for quality assurance of dosimetry calculations in radiotherapy. IAEA, Jan 2008. 5 Ojala J, ‘‘The accuracy of the Acuros XB algorithm in external beam radiotherapy A comprehensive review’’. Int J Cancer Ther Oncol 2014; 2:020417.

O028 Commissioning and clinical implementation of raystation v4.0.3.4 electron Monte Carlo D. Loria, B. Mzenda, R. Sims, D. Hutton Auckland Radiation Oncology. ([email protected]), ([email protected]), ([email protected]), ([email protected]) Currently, our electron patients are treated based on the square root cutout factor method in RadCalc. This method predicts output factors of the rectangular fields from the measured square field output factors. Although this method was shown to be accurate for rectangular fields1, this is not CT-based to account for density variations or

O029 Commissioning of the electron module in the pinnacle3 treatment planning system J. Baines, M. Chan Radiation Oncology Mater Centre, Princess Alexandra Hospital, Brisbane, Queensland Introduction The Pinnacle3 treatment planning system (TPS) uses a three-dimensional dose calculation for electron beams based on the Hogstrom [1] pencil beam algorithm and the limitations of this algorithm are well-documented. The aim of this work was to investigate the potential improvements to the accuracy of calculated output factors (OFs) that may be achieved using the methodology of McNutt [2] that is incorporated in the TPS electron module. Method Commissioning of the electron module requires square field OFs to be measured over the range of clinically used energies, applicators, field sizes and source-to-surface distances (SSDs). Factors were measured for a Varian linear accelerator using 6, 9, 12, 15 and 18 MeV electron beams and various square field sizes within 6, 10, 15 and 20 cm applicators at both 100 cm and 110 cm SSD. Pinnacle3-derived square and rectangular OFs were compared to corresponding measured OFs as well as those calculated using an inhouse Clarkson-based clinical methodology [3]. Additionally, Pinnacle3 OFs for a sample (n = 90) of electron treatments at 100 cm SSD were compared to in-house calculated OFs. Results Discrepancies of similar magnitude (B 1.5 %) were found between Pinnacle3, measured and in-house output factors at 100 cm SSD and the greatest differences occur at 110 cm SSD (4.5 %). Inhouse and Pinnacle3 OFs for electron treatments at 100 cm SSD exhibited discrepancies of ± 8 %.

123

Australas Phys Eng Sci Med Conclusion Differences between Pinnacle3 and in-house calculated OFs may be attributed in part to the contour of a patient which features in the TPS calculation but not in the in-house program, the latter assuming a flat surface. Experimental simulations of electron breast boost fields suggest that Pinnacle3 computed MU values may be more reliable for such cases and improve the treatment prescription accuracy relative to the current in-house methodology.

O031 Extension of QA for cranial stereotactic radiosurgery from cone based to HD-MLC based treatments R. Artschan1, R. Jones1, P. Ostwald1, P. Pichler1, J. Simpson1, J. Lehmann1,2 1

References 1. K. R. Hogstrom et al. (1981) Phys. Med. Biol. 26(3): 445–459. 2. T. R. McNutt and W. A. Tome´ (2002) Med. Dosim. 27(3): 209–213. 3. F. M. Khan (2003) The Physics of Radiation Therapy, 3rd edition, Philadelphia, PA: Lippincott Williams & Wilkins.

O030 Optimisation of ETB3 dosimetry for quality assurance of SRS and hypo-fractionated VMAT radiotherapy treatments C. E. Jones1, K. Y. Biggerstaff1, S. Ibrahim1, P. H. Charles1,2 1

Princess Alexandra Hospital, Brisbane, QLD, Australia. ([email protected]). 2Science and Engineering Faculty, Queensland University of Technology, Brisbane, Australia Introduction For high spatial resolution EBT3 film is a useful tool for patient specific dosimetry of SRS and SABR patients. However the maximum doses per fraction are high (typically between 5 and 18 Gy, but up to 40 Gy). Here we investigate the optimum RGB channel to use for the dose range of interest, and the effectiveness of custom implementations of dose response correction techniques: based on digitization of pre-irradiated film; and dual and multi-channel dosimetry1. Method Three delivered dose (0–50 Gy) versus EBT3 response calibration data sets were measured simultaneously by irradiating three stacked calibration squares per delivered dose point in plastic water. The calibration squares were scanned pre- and post-irradiation on an EPSON XL10000 flat bed scanner. Measured dose errors were calculated by applying the calibration curve (a linear interpolation of the delivered dose versus EBT3 response) from one calibration data set to the EBT3 response of the remaining two data sets. Results The dose errors from the additional two measured calibration sets varied: for the green channel, the (mean, stdv, max) errors in the 5–40 Gy dose range were: 1.2 %, 1.5 %, 3.0 %, compared to 2.9 %, 1.4 %, 4.8 %. Correction based on a scan of the pre-irradiated film did not consistently improve accuracy for the red or green channel. In the dose range of 0.5–10 Gy, a response correction technique based on the red and green channel decreased dose estimation errors (mean, stdv, max) for the combined additional 2 data sets to: 1.0 %, 0.8 %, 2.5 %. Conclusion Based on the dose estimation errors observed in this study, absolute dose dose can be measured with an accuracy of 5 % with the green channel in the 5–40 Gy dose range. In a limited dose range, using correction techniques, this uncertainty may be reduced. References 1. Micke, A., Lewis, D.F., and Yu, X (2011) Multichannel film dosimetry with nonuniformity correction, Medical Physics, 38(5), pp 2523–34

123

Calvary Mater Newcastle, Newcastle, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]). 2The University of Sydney, Sydney, Australia. ([email protected]) Introduction Following the clinical introduction of a new linear accelerator (Varian TrueBeam with HD-MLC) our cranial stereotactic radiosurgery program was extended from Varian Trilogy cone-based treatments. Prior to commencing HD-MLC Eclipse based treatments, planning and QA implications were assessed. This work reports on QA findings. Method Representative patient cases were investigated: (1) single 1 cm lesion, (2) single 2.5 cm lesion and (3) 3 lesions: 0.8, 0.9 and 1.1 cm. For cases 1 and 2, 69 and 6FFF Eclipse plans (AAA and AXB) were produced including both coplanar (CO) and non-coplanar (NC) aperture based and VMAT techniques. Using a Lucy phantomTM, dose was measured isocentrically with a CC04 ion-chamber and for selected planes with EBT3. Small field correction factors were not applied. For case 3, a single-isocentre CO VMAT 6 9 plan was investigated with EBT3 film. Results Consistently 1–2 % higher isocentre doses were calculated with AXB compared to AAA. Ion-chamber doses were within 3 % of total plan doses and within 5 % per field, with the exception of a VMAT NC field, which exhibited especially increased MLC shielding of the chamber. Almost all case 1 ion-chamber and film measurements were lower than planned (inconsistent with case 2), attributed to beam modelling combined with known ion-chamber limitations for small fields. Case 2 film agreement was good with isocentre doses within 2 % for all four plans. Case 3 single-isocentre multiple-lesion film measurements highlighted a modelling issue with disagreement of up to 15 % in high dose regions. Conclusion Ion-chamber suitability was limited due to lesion size (case 1) and MLC involvement. Results were good overall for the larger single lesion case. Disagreement between plan and measurement worsened with high modulation and small apertures and was unacceptably large in the case of single-isocentre multiple-lesions, highlighting the importance of testing the beam model under a variety of clinical conditions.

O032 Verification of complex treatment plans with a rectilinear end-to-end audit phantom for stereotactic radiotherapy R. Jones1, R. Artschan1, P. Ostwald, J. Lehmann1,2 1 Calvary Mater Newcastle, Newcastle, Australia. ([email protected]), ([email protected]), ([email protected]). 2Institute of Medical Physics, The University of Sydney, Sydney, Australia. ([email protected])

Australas Phys Eng Sci Med Introduction Phantoms currently available specialised for verification of stereotactic radiation therapy treatments are tailored towards measurement of small fields centred at isocentre. Using such phantoms for the implementation of multi-target single-isocentre treatments may require phantom shifts, multiple measurements and other complicating steps and an alternative design may be more suited. A previously introduced End-to-End audit phantom, rectilinear shaped for ease of use, was investigated for general suitability for small field verification. Measurements of simple and complex treatment plans were compared to calculations with two algorithms and also to measurements with a commercial spherical phantom, which is the current standard for QA. Method Initially, simple plans consisting of single square MLC and jaw shaped small fields (0.5–4.0 cm) were delivered to radiochromic film in the novel rectilinear phantom. Established film analysis procedures were employed and results compared with Eclipse calculations. Based on previous experience and literature, the Lucite relative electron density was overwritten to 1.151. To complement the basic plan delivery, a set of clinically representative VMAT plans were delivered to both phantoms. Results The initial simple field measurement results confirmed the viability of the rectilinear phantom for use as a small field verification tool. Measurements performed over multiple days were consistent with each other. The agreement between measurements and calculations of the clinical VMAT plans using the rectilinear phantom was within 0.6 % of that found using the commercially available phantom for coplanar plans and 3.9 % for non-coplanar plans. The agreement between phantoms was improved using the ACUROS algorithm compared to the AAA algorithm which may be due to inhomogeneity within the rectilinear phantom, for which density overrides were applied where appropriate. Conclusion Rectilinear phantoms can be used with similar accuracy as spherical phantoms under the scenarios tested. This opens the way for the development of new phantoms for quality assurance testing of multi-target plans single-isocentre plans. DVH comparison with different leaf constraints:

O033 A survey of modulated therapy use in Australia & New Zealand J. Barber1, P. Vial2 1

Nepean Cancer Care Centre, Sydney, Australia. ([email protected]). 2Liverpool and Macarthur Cancer Therapy Centres, Sydney, Australia. ([email protected]) Introduction Commissioning and quality assurance (QA) of intensity-modulated radiation therapies is a complex task, with many variables in the equipment and methods used to configure and verify treatment planning and delivery systems. In Australasia, Radiation Oncology Medical Physicists (ROMPs) manage this task in the absence of national guidelines and typically with minimal knowledge sharing across departments in our region with the same equipment. A comprehensive survey of Australasian departments was undertaken to capture a snapshot of current usage, commissioning and QA practices for intensity-modulated therapies. Method An online survey was developed and advertised to the ACPSEM ROMP membership, requesting a single response from each department. The survey consisted of 147 questions in total, covering IMRT, VMAT and Tomotherapy, and details specific to different treatment planning systems. Questions captured detailed information on equipment, policies and procedures for the commissioning and QA of each treatment technique. Results 42 responses were collected, representing 59 departments out of the 78 departments operational. 137 and 84 linacs from these departments are using IMRT and VMAT respectively, from a total 150 linacs. 4 Tomotherapy linacs and 1 Cyberknife are operating. All but 1 respondent is treating with IMRT. Over half have completed more than 500 IMRT treatments or have more than 5 years experience. About 2/3 of respondents are treating with VMAT. Of those treating with VMAT, half have less than 1 year clinical experience. There are at least 8 different treatment planning systems being used for IMRT or VMAT. There are large variations in all aspects of QA policies and procedures. Conclusion IMRT is almost universally implemented in the region. VMAT is being rapidly adopted, with about 1/3 of centres having implemented VMAT in the 12 months prior to the survey. Large variations in practice highlight the need for national guidelines on IMRT and VMAT.

O034 Social factors influencing participation in audits I. M. Williams Australian Clinical Dosimetry Service, Yallambie, Vic, Australia. ([email protected]) Introduction The Australian Clinical Dosimetry Service, (ACDS) is a national independent dosimetric audit service which has matured from blueprint to a functioning program. A major component of the ACDS success is an active interest and engagement with the ACDS audit program by a variety of stakeholders: regulators, professional colleges, clinical staff, patient representatives and Health Departments. This presentation draws on international examples to investigate why such a broad range of stakeholders are investing the

123

Australas Phys Eng Sci Med future of the ACDS. This question is especially relevant as the ACDS transfers from a fully funded to a user-pays funding model. Method The Michigan Keystone Project1 is an example of a successful quality improvement program in the clinical space. Analysis2 has provided six key factors which influenced the project’s success. The ACDS program is reviewed against these six factors: 1. 2. 3. 4.

Social (Isomorphic) Pressures Networked community efforts Re-framing dosimetry issues partly as a non-technical problem Changing practice and culture at the ‘sharp end’ by using interventions with different effects 5. Using large data 6. Skilfully using ‘Hard Edges’. Results All six factors identified above, frequently in concert, are influential to ACDS stakeholder engagement. The ACDS data for 1 & 2 is drawn from post-audit surveys while 6 is obtained form the ACDS audit records. Anecdote and a small number of examples inform the ACDS about stakeholders are influenced by 3, 5 and 6. Without a more rigorous approach to obtaining the relevant data, it is difficult to be more specific. Conclusion As the ACDS moves into the user-pays environment it is essential that it maintains and strengthens the attraction to participate in its audit program. The trust it has built within the stakeholder community must be maintained as it enhances the audit program to incorporate IMRT, FFF, Small Field and Dynamic Arc treatments. References 1. Provonost et al., An intervention to reduce catheter-related bloodstream infections in the ICU. New England Journal of Medicine, 355(26):2725, (2006). 2. Dixon-woods M, Bosk CL, Aveling EL et at., Explaining Michigan: Developing an Ex Post Theory of a Quality Improvement Program. The Milbank Quarterly, 89(2), 167–205, (2011).

O035 Credentialing for Stereotactic Ablative Body Radiotherapy (SABR) trials using a small mailable phantom T. Kron1, P. Lonski1, M. Crain2, B. Dixon1, S. Siva1 1

Peter MacCallum Cancer Centre, Melbourne, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]). 2TROG Cancer Research, Newcastle, Australia. ([email protected]) Introduction Credentialing of centres participating in radiotherapy clinical trials has become standard practice when introducing advanced technology and techniques. SABR is one of these techniques and several clinical trials are open or are under development for SABR in lung lesions. We aim to increase the efficiency of multicentre lung SABR credentialing through development of a compact phantom that can be mailed to participating centres. Method A phantom was designed from water equivalent material (CIRS) with an inclusion of cork mimicking lung. The phantom with total dimensions of 20 9 10 9 12 cm3 (2 kg) is shown in Fig. 1. It consists of two halves with a sheet of film sandwiched in between. The film allows probing the dose distribution perpendicular to a small field of radiation both in water and lung equivalent material. Three cm below the film plane are holders for thermoluminescence dosimeters (TLDs) and/or an ionisation chamber (IBA CC13). Results Figure 1b shows the dose distribution in the phantom calculated with Varian Eclipse Analytic Anisotropic Algorithm (AAA)

123

Fig. 1 a Photo of the phantom, b CT image with dose calculation for two small fields comparing water and lung equivalent inserts, c) close up of the cork

for two fields of 3 9 3 cm2 using 6 MV X-rays with the perturbation due to lung clearly visible. The phantom is of modular design with removable cylinders pre-loaded with TLDs and radiochromic film. By changing the TLD holders participating centres are asked to irradiate the phantom initially using a reference field of 10 9 10 cm2, followed by the small field irradiation as shown in the figure. Comparison of measurements and the treatment plan provided by the centre allows assessment of small field output factors and accuracy of the planning system in modelling small fields. Conclusion The phantom designed for credentialing of centres participating in clinical trials of lung SABR is currently undergoing field testing and is planned to be clinically available in late 2015.

O036 Gender diversity in our workplace: why does it matter? Natalka Suchowerska1, Roksolana Suchowerska2 1

Chris O’Brien Lifehouse, NSW. ([email protected]). School of Social and Political Sciences, University of Melbourne, Melbourne, VIC, Australia. ([email protected])

2

As medical physicists and biomedical engineers, our capacity to solve problems and engage in quality management plays a big role in enabling us to provide the highest standard of medical care. Building teams of likeminded people can give a sense of safety (particularly in our niche area), however it can also constrain the diversity of perspectives and skills (whether technical or interpersonal) that determines how we interpret and solve problems and how effectively we collaborate with the broad range of specialists that rely on us. This paper introduces the session on gender diversity by arguing that promoting diversity in our workplace is critical not just to uphold social standards and expectations about equal employment opportunity, but also to enable teams of medical physicists and biomedical engineers to have the capacity to provide the highest standard of medical care. By considering the unique contribution that women make to the workplace, the paper emphasises that gender diversity needs to be examined and promoted at all levels (from entry to management).

Australas Phys Eng Sci Med

O037 Women in medical physics and biomedical engineering: the current status of female ACPSEM members in Australia and New Zealand

O038 Women in medical physics: numbers in Australasia S. B. Crowe1, T. Kairn2

1

2

Eva Bezak , Roksolana Suchowerska , Elizabeth Claridge Mackonis3, Anna Ralston4, Lynne Greig5, Annette Haworth6, Tanya Kairn7,8, Natalka Suchowerska3 1 School of Physical Sciences, University of Adelaide, Adelaide, SA, Australia. ([email protected]). 2School of Social and Political Sciences, University of Melbourne, Melbourne, VIC, Australia. ([email protected]). 3Chris O’Brien Lifehouse, NSW, Australia. ([email protected]), ([email protected]). 4St George Hospital Cancer Care Centre, NSW, Australia. ([email protected]). 5 Wellington Blood and Cancer Centre, Wellington, New Zealand. ([email protected]). 6Peter MacCallum Cancer Centre, Melbourne, VIC, Australia. ([email protected]). 7 Genesis Cancer Care Queensland, Brisbane, QLD, Australia. 8 Queensland University of Technology, Brisbane, QLD, Australia. ([email protected])

Introduction Recent data from the Australian Institute of Physics reports a preference by females for life sciences to physical sciences, with a 2:1 ratio at school and a 4:1 ratio at university [1]. This study investigates the current status of women in medical physics (MP) or biomedical engineering (BME) professions in Australia and New Zealand by examining the experiences, opportunities and challenges that women in our profession face. Methods A survey was designed to identify the demographics of ACPSEM female members and to determine their aspirations, motivations, workplace experiences. All 205 female members (30 % of total membership) were invited to complete the survey online. SPPS software [2] was used to analyze the data and identify dominant correlations. Results The 102 survey responses in conjunction with ACPSEM membership records show that *80 % of female members working in MP and BME are employed full time and *75 % are aged between 26 and 45. Primary challenges reported were achieving certification and managing work-life balance. The respondents agreed that having a supportive manager was important in fostering their desired worklife balance. Respondents were passionate about work (wishing to contribute to treatment, education and R&D) while financial reward emerged as a secondary motivator. Although ACPSEM membership indicates that the number and proportion of women has increased over the past two decades, our findings suggests that this is not yet reflected in leadership roles. Conclusions Whilst the number of women in the profession is growing, some responses suggest there is an ongoing need for changes in employment practices. There is scope to increase number of women in leadership roles. To encourage this, organisations could equip managers with tools to support team members (particularly women) in managing competing work-life responsibilities. Additionally, further research is required to understand the role of female leadership in supporting gender diversity in the workplace. References 1. Bell S (2014), Women in the Science Research Workforce, Australian Institute of Physics Congress, Canberra 2014 2. IBM Corp. Released 2013. IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp.

1

Royal Brisbane & Women’s Hospital, Brisbane, Australia. ([email protected]). 2Genesis Cancer Care Queensland, Brisbane, Australia. ([email protected]) Introduction The status of women within the field of medical physics has been the subject of recent investigation and discussion, with a survey carried out by the International Organization for Medical Physics (Tsapaki & Rehani, 2015) and the inclusion of the topic as a theme at the most recent IUPESM World Congress. This study looks at the current numbers in Australasia, and how they compare with both the wider science and physics community, and with the medical physics community internationally. Method Lists of medical physicists were made using the ACPSEM website, conference proceedings and journal volumes. Genders of physicists were assigned, where not known to the authors, via online profiles, census- and sociolinguistic-informed estimation and personal communication. Existing demographic data were also aggregated. Results Demographic data for the Australasian cohorts are presented in Tables 1 and 2. Conclusion The representation of women with medical physics in Australasia is similar to the worldwide figure (28 %). Female participation was smaller than across Europe (36 %), where the European Commission set a target of 40 % in 1999. While there is no apparent under-representation (relative to the workforce) in local conference participation, there is a larger discrepancy in senior positions and journal editorials. The gender imbalance in department head appointments is similar to the wider sciences, where, for example, women comprise only 17 % of professorial positions in universities.

Table 1 Workforce demographics Cohort

N

Male (%)

Female (%)

2008 ROMP workforce survey

209

67

33

2015 Registered ROMPs

233

71

29

2015 Department head ROMPs

38

79

21

Table 2 Research demographics Cohort

N

Male (%)

Female (%)

2001–2014 APESM editorials

45

87

13

2010–2013 EPSM speakers

591

73

27

2010–2013 EPSM invited speakers

11

63

36

2010–2013 EPSM keynote speakers

20

75

25

References 1. Tsapaki, V & Rehani, MM (2015) Female medical physicists: The results of a survey carried out by the International Organization for Medical Physics. Phys Med 31: 368–373

123

Australas Phys Eng Sci Med

O039 Clinical Medical Physicist Gender Balance in Australasia

KS05 Development of a Total-Body Positron Emission Tomography Scanner

W. H. Round

S. R. Cherry1, R. D. Badawi2

School of Engineering, University of Waikato, Hamilton, New Zealand. ([email protected])

1 Department of Biomedical Engineering, University of California, Davis, USA. ([email protected]). 2Department of Radiology, University of California, Davis, USA. ([email protected])

Clinical medical physics is a male-dominated profession. But the situation is changing quickly. A survey of older members of the profession has shown that, as expected, 30 years ago there were very few clinical medical physicists in Australia and New Zealand. The proportion of female medical physicists has changed considerably since then with about 30 % now being female. A major factor that limits the growth of the female cohort is the low proportion of physics graduates who are female. Over the last five years in New Zealand this has been consistently about 27 %, but this percentage is unlikely to increase. However, the number of physics graduates has increased over the last 10 years. As registrar recruitment in New Zealand is highly coordinated by the hospitals acting collectively, good recruitment data is available and it is possible to analyse by gender the proportion of applicants who are successful. It seems from an initial analysis that a female graduate applying for a position in a hospital as a registrar is about 50 % more likely to be successful than a male.

KS04 Prospective risk analysis in radiation medicine T. Pawlicki Department of Radiation Medicine & Applied Sciences, Univ of California San Diego, USA. ([email protected]) The use of radiation in medicine occurs within a complex sociotechnical system that is heavily reliant on human operators. This, together with the fact the radiation is inherently dangerous, has led to a long history of focusing on safety in radiation medicine. One strategy to improve safety is found in high reliability industries, namely, to prospectively address issues of safety (hazards) rather than only improving safety after a near-miss or accident that reaches the patient. In the upcoming TG-100 report, the American Association of Physicists in Medicine (AAPM) has, in part, highlighted the tool of Failure Modes and Effects Analysis (FMEA) as a method for prospective hazard analysis. FMEA is a hazard analysis technique that has its roots in reliability theory meaning that components (i.e., equipment, process steps, or people) are assumed to fail with some known probability. Other industries, such as aviation, have also approached the problem of safety using hazard analysis techniques based on systems theory and treats safety as a problem of system control. This presentation will provide an explanation of FMEA hazard analysis technique with an example from the radiation oncology literature. Similarly, the System-Theoretic Process Analysis (STPA) hazard analysis technique, will also be described using an example. While no hazard analysis technique is ideal for all situations or environments, there are some clear differences between FMEA and STPA that will be elucidated. Both methods will be compared and contrasted to provide a better understanding of when one method might be preferred over the other. Whatever the hazard analysis technique used, it is important that safety improvement solutions are tested and deemed appropriate if they are to be successfully used across different centers.

123

Positron emission tomography (PET) is the highest sensitivity technique for human whole-body imaging studies. However, current clinical PET scanners do not make full use of the available signal, as they only permit imaging of a 15–25 cm segment of the body at one time. Given the limited sensitive region, whole-body imaging with clinical PET scanners requires relatively long scan times and subjects the patient to higher than necessary radiation doses. The EXPLORER initiative aims to build a 2-meter axial length PET scanner to allow imaging the entire subject at once, capturing nearly the entire available PET signal. EXPLORER will acquire data with *40-fold greater sensitivity leading to a six-fold increase in reconstructed signal-tonoise ratio for imaging the total body. Alternatively, total-body images with the EXPLORER scanner will be able to be acquired in *30 s or with *0.15 mSv injected dose, while maintaining current PET image quality. The superior sensitivity will open many new avenues for biomedical research. Specifically for cancer applications, high sensitivity PET will enable detection of smaller lesions. Additionally, greater sensitivity will allow imaging out to 10 half-lives of positron emitting radiotracers. This will enable (1) metabolic ultrastaging with FDG by extending the uptake and clearance time to 3–5 h to significantly improve contrast and (2) improved kinetic imaging with short-lived radioisotopes such as C-11, crucial for drug development studies. Frequent imaging studies of the same subject to study disease progression or to track response to therapy will be possible with the low dose capabilities of the EXPLORER scanner. The low dose capabilities will also open up new imaging possibilities in pediatrics and adolescents to better study developmental disorders. This talk will review the basis for developing total-body PET, potential applications, and review progress to date in developing EXPLORER, the first total-body PET scanner.

IS04 MARS spectral imaging: A new tool for radiology Prof. Anthony Butler1,2 1

Department of Radiology, University of Otago, New Zealand. Department of Physics and Astronomy, University of Canterbury, New Zealand. ([email protected]). On Behalf of the MARS research team: www.bioengineering.otago.ac/mars

2

The goal of the MARS project is to improve radiological diagnosis by providing molecular (functional) information [1]. To date the team has built a small bore (10 cm) scanner to investigate potential applications of spectral imaging [2,3]. The MARS system incorporates the Medipix3 spectral photon counting detector on a rotating gantry. The team has developed novel spectral reconstruction methods, quantification methods, and data visualization tools. In 2014, the team received funding to build a prototype human scanner to enable human clinical trials. The medical imaging applications we are studying include: noninvasive assessment of bone and cartilage health; measuring proliferating blood vessels in cancer, and imaging nanoparticles specifically

Australas Phys Eng Sci Med targeted to cancer; and non-invasively determining atheroma formation and stability. References 1. Clinical applications of spectral molecular imaging: potential and challenges. NG Anderson, AP Butler. Contrast media & molecular imaging 9 (1), 3–12 2. MARS spectral molecular imaging of lamb tissue: data collection and image analysis Aamir R, Chernoglazov A, Bateman CJ, Butler, P H Butler APH, Anderson NG, Panta R, Healy J L, Mohr J L, Rajendran K, Walsh M F, de Ruiter N J, Gieseg S, Woodfield T, Renaud P F, Brooke L, Abdul-Majid S, Clyne M, Glendenning R, Bones P J, Billinghurst M, Bartneck C, Mandalika H, Grasset R, Schleich N, Scott N, Nik S J, Opie A, Janmale T, Tang D N, Kim D, Doesburg R M, Zainon R, Ronaldson J P, Cook N J, Smithies D J, Hodge K. Accepted to JINST. 10.1088/1748-0221/9/02/P02005 3. Vision 20/20: Single photon counting X-ray detectors in medical imaging. Katsuyuki Taguchi and Jan S. Iwanczyk. Med. Phys. 40, 100901 (2013). doi:10.1118/1.4820371

O040 Registration of in vivo mpMRI with histology to validate tumour location and tumour characteristics in the prostate

freeform deformable image registration (DIR) method by Rueckart et al.2 Corresponding feature points and dice coefficients values were assessed to validate the in vivo to ex vivo registration. Results Overall mean error between in vivo MRI and histology was 3.2 mm, which included an average error between in vivo and ex vivo MRI of 3.1 mm after DIR was applied. Dice coefficients for the prostate contour increased from 0.83 to 0.93 after DIR, and from 0.69 to 0.75 for the peripheral zone contour. Conclusion This proof-of-concept study has demonstrated our registration framework provides a feasible method to map in vivo mpMRI with histology. Ongoing work includes processing to reduce image distortions and co-registration of cell density and tumour location maps from histology3, 4, ADC maps from DWI, pharmacokinetic maps from DCE-MRI and R2* maps from BOLD MRI. Acknowledgements Dr Reynolds is funded by a Movember Young Investigator Grant awarded through Prostate Cancer Foundation of Australia’s Research Program. Dr Williams was partially supported by a Victorian Cancer Agency Fellowship. Work was supported by PdCCRS grant 628592 with funding partners: Prostate Cancer Foundation of Australia and the Radiation Oncology Section of the Australian Government of Health and Aging and Cancer Australia. Thanks to Dr C Meyers for assistance in initiating this project, Prof Martin Ebert for his contribution during project development and Dr Rajib Chakravorty for his work towards the registration method development. References

1,2

2,3

4

4

H. M. Reynolds , S. Williams , A. Zhang , D. Rawlinson , C. S. Ong5, M. Esteva6, C. Mitchell7, B. Parameswaran3, M. Finnegan1, G. P. Liney8, Y. Sun1,2, A. Haworth1,2 1

Department of Physical Sciences, Peter MacCallum Cancer Centre, Melbourne, Australia. ([email protected]). 2Sir Peter MacCallum Department of Oncology, University of Melbourne, Melbourne, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]). 3Division of Radiation Oncology and Cancer Imaging, Peter MacCallum Cancer Centre, Melbourne, Australia. ([email protected]). 4Electrical & Electronic Engineering, University of Melbourne, Melbourne, Australia. ([email protected]), ([email protected]). 5Machine Learning Research Group, NICTA, Canberra, Australia. ([email protected]). 6 Biomedical Engineering, University of Melbourne, Melbourne, Australia. ([email protected]). 7Department of Pathology, Peter MacCallum Cancer Centre, Melbourne, Australia. ([email protected]). 8Ingham Institute for Applied Medical Research, Liverpool Hospital, NSW, Australia. ([email protected]) Introduction We have developed a registration framework to map multiparametric MRI (mpMRI) with ‘ground truth’ histology data of the prostate, to validate tumour location and tumour characteristics. Our goal is to use mpMRI data from a growing database of patients, to plan biofocussed prostate brachytherapy according to the radiobiological model validated by Haworth et al.1 Treatment plans will be designed to give a higher radiation dose to tumours and a lower dose to benign prostate tissue. Method In vivo mpMRI data was obtained from 6 patients scheduled for radical prostatectomy. After surgery, prostate specimens were formalin fixed, placed in a custom-made sectioning box and ex vivo MRI scanned. Tissue blocks of 5 mm thickness were cut and histology sections microtomed. Histology was registered with ex vivo MRI via a similarity transform using matching control points. In vivo and ex vivo 3D T2w MR images were convolved with a Laplacian of Gaussian filter to extract dominant features and registered using a

1. A. Haworth et al., ‘‘Validation of a radiobiological model for lowdose-rate prostate boost focal therapy treatment planning.,’’ Brachytherapy 12(6), 628–636 (2013). 2. D. Rueckert, L.I. Sonoda, C. Hayes, D.L. Hill, M.O. Leach, and D.J. Hawkes, ‘‘Nonrigid registration using free-form deformations: application to breast MR images,’’ IEEE Trans. Med. Imaging 18(8), 712–21 (1999). 3. H.M. Reynolds et al., ‘‘Cell density in prostate histopathology images as a measure of tumor distribution,’’ in Proc. SPIE 9041, edited by M.N. Gurcan and A. Madabhushi (2014). 4. M.D. Difranco, H.M. Reynolds, C. Mitchell, S. Williams, P. Allan, and A. Haworth, ‘‘Performance assessment of automated tissue characterization for prostate H & E stained histopathology,’’ in Proc. SPIE 9420, edited by M.N. Gurcan and A. Madabhushi (2015).

O041 Assessment of MRI sequences for imaging implantable radiotherapy markers in the prostate Robba Rai1, Jason Dowling2, Sankar Arumugam3, Benjamin Schmitt4, Aitang Xing3, Lois Holloway3,5, Gary Liney3,5 1 Liverpool Cancer Therapy Centre, Liverpool, Australia. ([email protected]). 2CSIRO, Royal Brisbane and Women’s Hospital, QLD, Australia. ([email protected]). 3 Department of Medical Physics, Liverpool Hospital, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]). 4Healthcare Sector, Siemens Ltd, NSW, Australia. ([email protected]). 5Ingham Institute for Applied Medical Research, Liverpool, Australia

Introduction Fidicual marker insertion in the prostate is common practice before radiotherapy (RT) to improve target localisation and fusion between computed tomography (CT) and set-up images. With increased use of magnetic resonance imaging (MRI) in treatment

123

Australas Phys Eng Sci Med

Fig. 1 Left 3D overlay of artefacts using FLASH (red) compared to UTE (yellow). Right Comparison of gold seed position from UTE (red) to CT (white) planning, the MRI appearance of these markers needs to be carefully characterised. A gradient-echo sequence is often used to increase the conspicuity of markers due to enhanced susceptibility artefacts but at the cost of reduced accuracy. Artefacts are known to depend on material, orientation to the magnetic field, imaging sequence and echo time (TE). The purpose of this study is to evaluate the appearance of two commercial fidicual markers for various sequences. Method All imaging was conducted on a dedicated 3T MRI simulator used for RT planning. A phantom was constructed containing both gold (1.2 9 3 mm) and polymer (1 9 3 mm) markers (Fig. 1) and imaged with MRI and CT. Five MRI sequences were tested: the current clinical sequence (FLASH), turbo spin echo (TSE), turbo gradient spin echo (TGSE), and two prototype sequence versions of ultra short echo time imaging (UTE and PETRA, TE = 40&60 ls, respectively). The phantom was imaged with the seeds positioned parallel and perpendicular to the main field and measurements of marker dimension in two axes were taken. Results The marker-induced artefacts in the FLASH images were 15 times greater in total volume compared to the smallest observed (UTE). TGSE reduced seed artefacts to 3 9 5 mm (gold/perpendicular orientation) and 3.5 9 6.5 mm (polymer/perpendicular). UTE images exhibited the smallest artefact dimensions (2 9 3 mm for gold/parallel and 3 9 3 mm for polymer/perpendicular) and a subvoxel accuracy (mean 1.2 mm) compared to CT. Conclusion Wide variation in seed appearance has been observed with polymer markers producing a slightly larger artefact than gold. Combined contrast and UTE sequences in particular improved accuracy and may have potential in a dedicated RT prostate protocol.

methods, along with predictions of tumour characteristics, in biofocused prostate brachytherapy treatment, where non-uniform distributions of radiation are calculated based on our validated radiobiological model (Haworth, 2013). Method In vivo mpMRI data were collected from 16 patients before radical prostatectomy. Sequences included T2-weighted (T2w), diffusion-weighted (DWI) and dynamic contrast enhanced (DCE) imaging. In vivo mpMRI was registered with ‘ground truth’ histology. Prostate contours were delineated by a radiation oncologist and tumours annotated on histology by a pathologist. Five patients with minimal imaging artefacts were selected for analysis. A Gaussian kernel SVM (Chang, 2011) was trained and tested on different subsets using leave-one-out cross validation, with mpMRI intensities as features and histology information as true labels. Prediction accuracy, defined as the percentage of voxels correctly classified, was used to assess performance. Results An overall prediction accuracy of 80.83 % for the five patients was achieved (Fig. 1), which is comparable with other related studies (Wang, 2014). Whilst a Gaussian kernel is capable of

Fig. 1 Predicted tumour locations for a patient (blue predicted tumour location; green line true tumour location; dashed prostate contour)

O042 Predicting tumour location from multiparametric MRI for bio-focused prostate radiotherapy using kernel support vector machines Y. Sun1, H. Reynolds1,2, D. Wraith3, S. Williams2,4, C. Mitchell5, A. Haworth1,2 1 Dept. of Physical Sciences, Peter MacCallum Cancer Centre, Melbourne, Australia. 2The Sir Peter MacCallum Dept. of Oncology, The University of Melbourne, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]). 3Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, Australia. ([email protected]). 4Division of Radiation Oncology and Cancer Imaging, Peter MacCallum Cancer Centre, Melbourne, Australia. 5Dept. of Pathology, Peter MacCallum Cancer Centre, Melbourne, Australia. ([email protected])

Introduction We investigate the performance of a support vector machine (SVM) classifier to predict prostate tumour location using multi-parametric MRI (mpMRI) data. Our intention is to apply these

123

Fig. 2 Relative importance of sequences in predicting tumour location measured independently by mean decrease accuracy and mean decrease Gini coefficient

Australas Phys Eng Sci Med

References 1. Haworth, A. et al. (2013). Validation of a radiobiological model for low-dose-rate prostate boost focal therapy treatment planning. Brachytherapy, 12(6), 628–636. 2. Chang, CC. et al. (2011). LIBSVM : a library for support vector machines. ACM Transactions on Intelligent Systems and Technology, 2:27:1–27:27. 3. Wang, S. et al. (2014). Computer aided-diagnosis of prostate cancer on multiparametric MRI: a technical review of current research. Physics in Medicine and Biology, 57(12), 3833–3851.

O043 Testing the impact of radiotherapy treatment quality on disease progression for prostate cancer M. Marcello1,2, M. A. Ebert1,2, A. Haworth3,4, A. Steigler5, A. Kennedy1, M. Bulsara6, R. Kearvell1, D. J. Joseph1, J. W. Denham8 1 Department of Radiation Oncology, Sir Charles Gairdner Hospital, Western Australia, Australia. ([email protected]), ([email protected]). 2School of Physics, University of Western Australia, Western Australia, Australia. ([email protected]), ([email protected]). 3 Department of Physical Sciences Peter MacCallum Cancer Centre, Victoria, Australia. 4Sir Peter MacCallum Department of Oncology University of Melbourne, Victoria, Australia. ([email protected]). 5Prostate Cancer Trials Group, Faculty of Health, University of Newcastle, Callaghan, New South Wales, Australia. ([email protected]). 6Institute for Health Research, University of Notre Dame, Fremantle, Western Australia [email protected]. 7School of Surgery, University of Western Australia, Western Australia, Australia. ([email protected]). 8School of Medicine and Public Health, University of Newcastle, New South Wales, Australia. ([email protected])

Introduction Multicentre radiotherapy clinical trials incorporate quality assurance methods for ensuring consistent application of the trial protocol. The purpose of this study was to determine whether post-treatment measures of disease progression are affected by factors describing the quality of radiotherapy treatment. Method The TROG 03.04 RADAR trial tested the impact of duration of androgen deprivation on outcomes for prostate cancer patients receiving dose-escalated external beam radiation therapy. The trial incorporated a plan-review process, level III dosimetric intercomparison study and an assessment of the setup errors at each participating institution, from which variables suggestive of treatment quality were collected. Kaplan–Meier (KM) and Cox proportional

PSA progression curve according to variation from protocol PSA Progression Curve for Per-Patient % B or C Variations 100

Percent PSA Progression Free

separating complex data, it brings the risk of overfitting. Therefore, other kernels such as linear kernels are under investigation. Additional investigations showed that apparent diffusion coefficient (ADC), from DWI, achieved the highest importance for predicting tumour location measured independently by two metrics (Fig. 2). Hence our efforts will focus on producing high quality ADC maps which we predict will also correlate with tumour cellularity and tumour aggressiveness. Conclusion We present an SVM classifier to predict tumour location from mpMRI with promising results on a small sample of patients. Future work will include more patients and apply similar techniques to predict tumour characteristics for biological optimisation in prostate brachytherapy treatment planning.

80

60

40 20 P = 0.0015743 o ≥ 5.26% * <5.26% 0 0 2 4 6 8 Time Since End of RT (years)

Number of patients at risk 419 387 339 297 257 228 189 119 50 <5.26% 315 298 278 248 215 199 158 85 30 ≥5.26%

10

9 4

0 0

Fig. 1 KM curve showing PSA progression for patients with greater or less than the mean percentage of protocol variations (5.26 %) hazards statistics were employed to analyse patient data for associations between a number of quality-related variables and two patient outcomes; PSA progression (PSAP) and local composite progression (LC). Results The measure of the variation from protocol of each patient’s planned treatment was consistently associated with significantly increased disease progression, particularly PSAP. This effect was amplified by dose escalation. Patient’s accrued from centres showing smaller setup accuracy errors were found to experience more progression. This effect was amplified in the high-risk disease group. Increased dose difference between measured and planned dose at the centre of the prostate resulted in more progression, however this effect was confounded by patient protocol variations. Increased dose delivered to other points around the prostate were shown to reduce progression. Patients treated at centres with higher accrual experienced less progression. Conclusion Compliance with treatment protocol is beneficial in reducing treatment failure as indicated by the association between patient protocol variations and increased disease progression. The impact of setup accuracy suggests caution is required when reducing margins, especially in treatment of patients with high-risk disease. Centres with higher accrual numbers tend to focus more on minimising progression than treatment toxicity (Fig. 1).

O044 Irregular surface compensation versus ‘‘ultradynamic wedges’’ for Breast IMRT A. J. Williams, L. Andrews, J. Donaldson, M. Satterthwaite, N. Whitaker Wellington Blood and Cancer Centre (WBCC), Wellington, NZ. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected])

123

Australas Phys Eng Sci Med Introduction At the WBCC, we wish to develop our breast technique from one using 3DCRT to an IMRT based approach, with the aims of (a) improving planning and treatment efficiency, (b) improving dose homogeneity and (c) reducing the use of 18 MV. The breast is an ideal site for the use of irregular surface compensated (ISC) radiotherapy. The purpose of this study therefore was to evaluate plans created either using (a) 3DCRT (b) using the built-in ISC functionality of the Eclipse (Varian Medical Systems) treatment planning system and (c) using an ‘‘ultra-dynamic-wedge’’ (UDW), which is a two-dimensional dynamic wedge created with the MLCs (Hand, 2014). Methods Plans on eight patient datasets were created using the three methods. Data was collected on (1) the time taken to produce a clinically acceptable plan, (2) the MU required, (3) the standard deviation, V98 and D2cc of the dose to the whole breast PTV, 4) the energy used and 5) the ‘ease of use’ (a summed score of ease of (a) learning the technique, (b) getting an optimal dose distribution, (c) the ability to drive the system through prior knowledge). Results The table below summarises the results of the study. ISC Average time to plan in minutes (range)

UDW

3DCRT

27

(17–38)

22

(14–35)

73

(20–150) Average MU per field

182

128

Average standard deviation of dose to 3.0 % PTV

205

3.2 %

3.7 %

PTV V98 (% of PTV receiving 98 % of 91.7 % the prescribed dose)

91.8 %

90.0 %

PTV D2cc (Dose received by 2 cc of the PTV)

105.4 % 106.0 % 106.3 %

Ease of use ranking (order of preference)

3

1

2

Both ISC and UDW techniques could be used to create good quality plans with 6 MV only. However, the UDW technique was quicker to plan and needed less MU for equivalent dose distributions. In addition, UDW was preferred by planning staff because it utilised the planners’ knowledge of conventional planning to drive the system to obtain the desired result in an intuitive way. Conclusion The UDW planning technique is currently the preferred method but further work is planned to test the accuracy of delivery of the two methods before a final selection is made.

1

Peter MacCallum Cancer Centre, Melbourne, Australia. ([email protected]), ([email protected]), ([email protected]). 2RMIT University, Melbourne, Australia. ([email protected]), ([email protected]), ([email protected]) Introduction Anthropomorphic phantoms are an essential tool for introduction of new treatment techniques in radiotherapy. They mimic human shapes and tissue properties and allow for the incorporation of radiation measurement devices to verify in an end to end test that dose is delivered as planned. This work aimed to utilise additive manufacturing (‘3D printing’) to create phantom parts that not only accurately show the anatomy but also include recesses for radiation detectors. Method A commercial anthropomorphic phantom (CIRS Norfolk Virginia) was imaged using a diagnostic CT scanner (Philips Brilliance wide bore). The images were transferred to Matlab software and bone structures auto-contoured. Software was developed to convert the DICOM data to a steriolithographic (STL) file suitable for additive manufacture. Phantom slices were manufactured from a CT image of anatomy between two slices of the anthropomorphic phantom as seen in Figure 1. Recesses for thermoluminescence dosimeters (TLDs) and radiochromic film were incorporated into the design. Results Several slices were printed using different material combinations for the head of the phantom as seen in figure 1. The anatomical features such as bone and air cavities can be matched to the adjacent slices within submillimetre accuracy. While radiation detectors such as TLDs and film can easily be incorporated, work is ongoing to optimise the materials to mimic adjacent materials not only in terms of shape but also radiological properties. Depending on the mechanical properties of the materials a shell can be included in the manufacturing process as visible in the insert of the figure. Conclusion It is possible to print anatomically accurate slices of DICOM files that can be used within an anthropomorphic phantom to accommodate radiation detectors of choice. In principle these slices can also mimic anatomy with pathological changes such as tumours or other individual patient dependent anatomical features (Fig. 1).

References 1. Varian Medical Systems, Palo Alto, California, USA. 2. Hand A, Williams A, Investigation into the use of custom-generated bi-directional wedged fluence maps for forward planned IMRT breast treatments, MPEC & biennial radiotherapy physics conference, Glasgow, Sept. 2014.

O045 Extending the functionality of anthropomorphic phantoms for radiotherapy dosimetry using additive manufacture P. Lonski1, M. Leary2, C. Keller2, E. Kyriakou1, R. Franich2, T. Kron1

123

Fig. 1 Picture of an anthropomorphic phantom (CIRS) with two additional slices printed using additive manufacturing. The slices fit anatomically between two phantom slices but could also replace a slice of the anthropomorphic phantom

Australas Phys Eng Sci Med

O046 Investigation of materials and processes in additive manufacture for modelling different human tissues for radiotherapy phantoms T. Kron1, C. Keller2, P. Lonski1, R. Franich2, E. Kyriakou1, M. Leary2 1

shape and size which will allow for creation of realistic anthropomorphic phantoms. Conclusion Additive manufacture is a relatively new technology with great potential for the creation of tissue substitutes in radiation dosimetry.

Peter MacCallum Cancer Centre, Melbourne, Australia. ([email protected]), ([email protected]), ([email protected]). 2RMIT University, Melbourne, Australia. ([email protected]), ([email protected]), ([email protected])

O047 3D printing of air slab caps for correction free small field dosimetry

Introduction Tissue equivalent materials play an important role in commissioning and quality assurance of radiotherapy treatment units. We are investigating different materials available for additive manufacture (‘‘3D printing’’) for their suitability for dosimetric phantoms. Of particular interest was the use of several materials in combination at different patterns and spacing to customise CT number. Method Several materials used for additive manufacture at the Centre for Advanced Manufacturing at RMIT University were studied. The materials were used individually in small (3 9 3 9 3 cm3) samples and in combination with each other using different spacing and ratios between materials. All materials were studied for radiological tissue equivalence using a diagnostic CT scanner (140 kVp, Philips Wide Bore Brilliance) in axial scans with 1 mm slice thickness and spacing. Scans of the samples were performed in water where possible to create scatter and beam hardening conditions similar to in vivo. Figure 1a shows one of the set-ups and Fig. 1b a typical scan. Hounsfield CT numbers were assessed in regions of interest using Philips software as shown in Fig. 1b. Figure 1c shows a set of samples with different ratios of two materials, in this case varying the radius of the internal material in a series of cylinders. Results In total more than 20 materials and material combinations were tested. CT numbers ranged from -630 to approximately +150 HU, covering a large range of possible tissue substitutes. The lower end of the CT number spectrum was achieved by interspersing air and plastic structures on a sub-millimetre scale. Using commercially available 3D printing devices the materials can be formed in any

1

B. Perrett1, P. H. Charles1,2, T. Markwell1,3, T. Kairn1,4, S. Crowe1,5 Science and Engineering Faculty, Queensland University of Technology Brisbane. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]). 2 Princess Alexandra Hospital, Brisbane, Australia. 3Radiation Oncology, Mater Centre, Brisbane, Australia. 4Genesis Cancer Care Queensland, Brisbane, Australia. 5Royal Brisbane and Women’s Hospital, Brisbane, Australia Introduction Air cavities can be used to correct for a density dependent over-response by commercial diode detectors in small fields1. Caps containing air cavities have been validated experimentally for diode detectors in small fields2,3. This project involves the design and 3D printing of caps for the Stereotactic Field Diode. Two designs are considered—a sealed cap containing a sealed air cavity, and an unsealed cap which locates the air cavity at the face of the detector. Method Monte Carlo simulations of both cap designs were performed to determine optimal air slab dimensions. The cap material (ABS) was added to a second set of simulations in order to quantify any effect on the optimal air gap size. Various changes to parameters were also simulated—cap layer dimensions, cap density changes due to 3D printing, and unwanted air gaps were considered. Preliminary verifications of the dimensions and dosimetric effect of the printed sealed cap design were performed. Results For the sealed cap design the optimal air gap size was 0.6 mm, which increased to 1.0 mm when the ABS was simulated. The unsealed cap design had less variation, with a 0.4 mm air gap optimal in both situations. Significant influence factors include air trapped at the face of the detector, changes in density, and changes in layer thickness. An over-correction introduced by the caps in an experimental verification is likely due to an unwanted air pocket in the cap. Conclusion The unsealed cap is much easier to produce than the sealed cap, with less sensitivity to uncertainties in density and cap layer thickness, but at the cost of the need for a smaller air gap thickness (requiring more accurate 3D prints) and the need for waterproofing in use. Simultaneous prints of several caps were found to result in caps of superior reproducibility. References

Fig. 1 A variety of samples including 3D print materials and other typical phantom materials for radiotherapy. a samples on CT scanner couch, b CT image with CT number assessment in regions of interest, c a series of samples with varying ratios of two materials

1. Charles, P et al. (2013) Monte Carlo-based diode design for correction-less small field dosimetry. Phys. Med. Biol. 58(13) pp4501 2. Charles, P. et al. (2014) Design and experimental testing of air slab caps which convert commercial electron diodes into dual purpose, correction-free diodes for small field dosimetry, Med. Phys. 41 101701. 3. Underwood, T et al. (2015) Validation of a prototype DiodeAir for small field dosimetry. Phys. Med. Biol. 60(7) pp2939

123

Australas Phys Eng Sci Med

O048 Use of 3D printing to produce radiologically robust phantoms T. Kairn1,2, S. B. Crowe2,3, T. Markwell4 1 Genesis Cancer Care Queensland, Brisbane, Australia. 2Queensland University of Technology, Brisbane, Australia. ([email protected]), ([email protected]). 3Royal Brisbane & Women’s Hospital, Brisbane, Australia. 4Radiation Oncology, Mater Centre, Brisbane, Australia. ([email protected])

Introduction Three dimensional printing has allowed the inexpensive fabrication of models of anatomical structures for medical applications. This study investigated the potential for producing patient specific phantoms with tissue-equivalent densities for radiotherapy dosimetry. Method The SketchUp modelling software was used to design 3 cylinders that could into a CIRS 062 CT phantom, with 90, 50 and 30 % ABS infill densities. Two scale models (50 % and 100 %) of a human lung were prepared using a patient CT image dataset and the CERR, 3D Slicer, and Meshmixer software. The models were printed using the Da Vinci 3D printer. The geometric properties were verified against design dimensions by both physical measurement and CT imaging. The radiological properties were evaluated by CT imaging and downstream EBT3 film measurements for eight treatment beams (6 and 10 MV photons; 6 MV FFF photons; and 6, 9, 12, 16 and 20 MeV electrons). Results All cylindrical phantoms matched their planned dimensions to within 1 mm and produced transmission results similar to cylindrical inserts from commercial tissue-equivalent phantoms. The physical properties of all printed samples were unchanged after intensive irradiation over a period of 6 weeks. The density, electron density relative to water, and linear attenuation coefficient relative to water are presented below. Phantom

Density

Relative Electron Linear attenuation Density coefficient

Cylinder— 90 % ABS

1.06 ± 0.02 1.05 ± 0.03

1.04 ± 0.03

Cylinder— 50 % ABS

0.58 ± 0.02 0.57 ± 0.02

0.56 ± 0.02

Cylinder— 30 % ABS

0.36 ± 0.01 0.35 ± 0.01

0.34 ± 0.01

Lung—10 % ABS

0.17 ± 0.16 0.16 ± 0.16

0.15 ± 0.15

Conclusion Three dimensional printing is capable of producing geometrically accurate and radiologically robust lung- and tissueequivalent phantom material. Appropriate infill densities were determined for tumours and soft tissue (90 %) and lung (30–50 %).

O049 Fabrication of dosimetric and anatomical phantoms using 3D printing Paul Liu1,2, Natalka Suchowerska1,2, David McKenzie2 1 Chris O’Brien Lifehouse, Camperdown, NSW, Australia. ([email protected]). 2School of Physics, The University of Sydney, NSW, Australia

123

Fig. 1 a The cylindrical phantom in CAD and b the 3D printed phantom. c A CT scan of the spinal cord and d the 3D printed representation Aim The aim of this work is to demonstrate the feasibility of using 3D printing technology to create dosimetric and customized anatomical phantoms that would otherwise be difficult to manufacture. Method The cylindrical dosimetric phantom, created in CAD software, was customized to hold an array of scintillation dosimeters to validate arc and multi-directional treatments. The anatomical phantom was created using patient CT data, which was converted into a 3D model. The phantoms were printed using fused deposition with the UP 3D printer. Both phantoms were printed as hollow objects and filled with two types of material: dental wax and gel bolus. The resulting phantoms were CT scanned to determine their density and uniformity and irradiated with a therapeutic beam to assess their suitability. Results The 3D printed phantoms are shown in Fig. 1. CT scans of the printed and filled phantoms showed a uniform electron density with less than 2 % variance. Gel bolus yields a uniform phantom that is close to water in mass density (qgel = 1.07 g cm-3). Dosimetric characterisation of the cylindrical dosimetric phantom filled with gel bolus showed low angular dependence (\0.6 %) indicating that the scintillator was located accurately on the axis. The anatomical phantom provided an accurate surface representation of the CT image. However, the phantom density does not reflect that of the real anatomy. Conclusion The fabrication of radiotherapy phantoms using 3D printing using fused deposition fast and cost effective. Customized patient dosimetry can be achieved with 3D printing of anatomical phantoms, however, more sophisticated printing techniques will be required to achieve the desired mass and electron densities.

O050 Clinically implementation of automated 3Dprinting technique to introduce better fitted bolus on patients undergoing radiotherapy A. U. Yeo, M. Goharian, B. Subramanian, W. Ding Physics Department, Radiation Oncology Victoria, GenesisCare, VIC, Australia. ([email protected]) Introduction Inexpensive 3D-printers are now available in worldwide commercial markets. This work demonstrates how an

Australas Phys Eng Sci Med Ultimaker2 can be used in context of EBRT with use of both photons and electrons beams. Method Ultimaker2 uses polylactide (PLA) with a known atomic composition of C6H8O4 as the printing material. Physical and radiological properties of this material were characterised before performing dosimetric measurements to verify suitability of 3Dprinted materials for radiotherapy. Two geometries, a uniform slab (1 cm thickness) and a customised nose block for anthropomorphic phantom were made using the 3D-printer. For the former case, dosimetric results were compared to solid water via measurements using ion chamber and film; for the latter, clinical and non-clinical scenarios were planned through TPS (Pinnacle) for both photon and electron beams—the calculated results were compared to its measurements using TLD, 2D-diode, and film. Results Energy-dependent Zeff calculations for PLA using the method of Taylor et al. (Med Phy 2012) show that the discrepancy is within 2–5 % for Compton scattering dominant energy range (0.1–10 MeV photon beams). The physical density of 3D-printed materials was in a range of 1.05–1.09 g/cc. Dosimetric verification results with the flat geometry show differences of CAX dose within 0.5 % and wellmatched line profiles. TLD measurements with a clinical scenario with two lateral photon beams show dose differences within 4 % at the interface of nose-block and skin surface. 2D-diode and film measurements in non-clinical scenarios show a good agreement with its TPS-calculated dose distributions in terms of 2D-gamma analysis3mm/3% ([98 %). Conclusion This work demonstrated that 3D-printed bolus can be accurately modelled through TPS. A 3D-printing method could be clinically implemented to enable automatic fabrication process of complex bolus structures. This technique allows not only to improve treatment quality but also to minimise staff workload as well as patient involvement. References 1. Taylor, M. L. et al. (2012) Robust calculation of effective atomic numbers: The Auto-Zeff software. Med. Phy. 39, 1769–78

KS06 Dosimetry for the MRI linac Bas Raaymakers Department of Radiotherapy, UMC Utrecht, The Netherlands. ([email protected]) The development of hybrid MRI radiotherapy systems, in our case the 1.5 T MRI linac developed in collaboration with Elekta and Philips, led to new challenges for dosimetry. Both the radiotherapy as the MR imaging part of the system needs to be calibrated and commissioned, but also the hybrid part, that is, the MRI feed-back loop for treatment adaptation needs to be addressed. Our starting point is using the existing protocols for the current radiotherapy and radiology clinic, but procedures needed to be modified, reinvented or developed and tested to address specific features such as the absence of a light field, but also dedicated to the hybrid character. A major difference is the fact that dose delivery is performed in the presence of a magnetic field. This will affect the dose distribution, which warrants dedicated (new) QA procedures, but also the response of radiation detectors are typically affected by the (orientation to the) magnetic field. Developments on reference dosimetry, machine quality assurance, patient quality assurance and work flow for the MRI linac will be discussed.

IS05 Small field dosimetry: still not a small problem D. Thwaites Institute of Medical Physics, The University of Sydney, Australia Abstract not yet supplied.

O051 Very small field dosimetry for elekta infinity with agility MLC: diode design and field size reporting S. Ibrahim1, E. Inness1, B. Perrett2, S. Crowe2,3, P. H. Charles1,2 1

Princess Alexandra Hospital, Brisbane, Australia. ([email protected]), ([email protected]). 2Science and Engineering Faculty, Queensland University of Technology Brisbane Australia. ([email protected]), ([email protected]), ([email protected]). 3Royal Brisbane and Women’s Hospital, Brisbane, Australia Introduction The over-response of diodes at all small fields can be cancelled out by placing an air gap on the proximal end of a diode, and negate the need for any correction factors1. The aim was to experimentally measure the optimal air gap thickness required for Elekta linacs. In a previous study on Varian linacs, it was shown that reporting output factors versus effective measured field size showed a very large reduction in uncertainty when comparing different linacs, compared to when they were plotted against nominal field size2. This study was repeated on three Elekta linacs with Agility collimators. Method A custom-made air-cap was placed on top of a PTW60017 electron diode. Output factors were measured using 4 air gap thicknesses (0, 0.5, 1.0, 1.5 mm) and square field sizes of side length 5–50 mm. The detector sensitivity correction factor was calculated for each field size, observing which air thickness it was unity for all field sizes3. Output factors measured using the PTW60017 without an air gap were plotted against both effective measured and nominal field sizes. Results A 1 mm air gap was optimal on the Elekta machines, agreeing with previous values for a Varian linac3. The plot of output factors versus effective field size showed excellent agreement between linacs (nominal field size did not). There was a significant decrease in output factors for field sizes\10 mm for Elekta machines, compared to Varian2. The difference is most likely due to greater source occlusion on the Elekta linacs. Conclusion The optimal air gap required to make a correction-free PTW60017 electron diode was 1.0 mm, similar to previous studies using the Varian iX. Reporting output factors as a function of effective measured field size greatly increases inter-machine comparability for the same machine type. However it was found that this method can not be used to cross-compare different machine designs. References 1. Charles, P. et al. (2013) Monte Carlo-based diode design for correction-less small field dosimetry. Phys. Med. Biol. 58 pp4501. 2. Cranmer-Sargison, G. et al. (2013) A methodological approach to reporting corrected small field relative outputs. Radiother. Oncol. 109 pp350.

123

Australas Phys Eng Sci Med 3. Charles, P. et al. (2014) Design and experimental testing of air slab caps which convert commercial electron diodes into dual purpose, correction-free diodes for small field dosimetry, Med. Phys. 41 101701.

O052 Applications of dose-area product in small field dosimetry using a large-area parallel-plate ionization chamber T. Kupfer1, J. Lehmann1, R. Franich1, D. Butler2, G. Ramanathan2, K. Rykers3 1 School of Applied Sciences, RMIT University, Melbourne, Australia. ([email protected]). 2Australian Radiation Protection and Nuclear Safety Agency, Yallambie, Australia. 3Austin Health, Radiation Oncology Centre, Heidelberg, Australia

Introduction A previously commissioned large-area parallel-plate chamber (LAC) (PTW34080, PTW Freiburg, Freiburg, Germany, 8.16 cm diameter) was used to further investigate novel applications of dose-area products (DAP) in stereotactic photon beams (1–4). Method DAP vs field size was measured in clinical Elekta 6 MV beams and compared to integrated dose calculated by a treatment planning system (TPS). DAPR20,10, the ratio of DAP at 90 and 80 cm SSD and a constant 100 cm source-detector distance was measured for 6 MV beams of collimated with 5, 7.5, 10 and 15 mm diameter cones. The LAC’s absolute dose response (cGy cm2 nC-1) was measured in 6 MV beams collimated by MLC and jaws (3 9 3, 5 9 5 & 10 9 10 cm) and a stereotactic cone (5 cm). Results There was very good agreement with TPS data when normalized to 5 9 5 cm field size: \0.2 % difference for 4 9 4, 3 9 3 and 2 9 2 cm. At 1 9 1 cm the difference was +1.2 % and the chamber signal was equally composed of primary beam and leakage. It was found that DAP is very sensitive to collimator accuracy. DAPR20,10 increased with decreasing field size from 0.658 ± 0.002 (15 mm cone) to 0.664 ± 0.002 (5 mm cone). This continues and amplifies a slight trend observed previously in DAPR20,10 for MLCdefined fields down to 1 9 1 cm. Absolute dose–response (in units of mGy cm2 nC-1) increases by almost 10 % in a 10 9 10 cm field compared to field sizes B 5 9 5 cm. Conclusion Measured DAP is useful to evaluate TPS data, however, collimator leakage can significantly contribute to the LAC’s signal in small fields. An increase of DAPR20,10 for very small field sizes indicates beam hardening effect, which is also reported on elsewhere and may be attributable to partial source occlusion.(6,7) Due to its over-response in large fields, the LAC should be calibrated in small fields that do not extend laterally beyond the chambers sensitive volume, preferably cones of fixed collimator size.

[updated 19 Nov 2010; cited 21 Oct 2013]. Available from: http://www.npl.co.uk/publications/science-posters/application-ofdose-area-product-and-dap-ratio-to-dosimetry-in-imrt-and-smallfield-external-beam-radiotherapy. 3. Sa´nchez-Doblado F, Hartmann GH, Pena J, Rosello´ JV, Russiello G, Gonzalez-Castan˜o DM. A new method for output factor determination in MLC shaped narrow beams. Physica Medica. 2007;23(2):58–66. 4. Djouguela A, Harder D, Kollhoff R, Ru¨hmann A, Willborn KC, Poppe B. The dose-area product, a new parameter for the dosimetry of narrow photon beams. Zeitschrift fur Medizinische Physik. 2006;16(3):217–27. 5. Duane S. The use of dose area product in beam quality specification for IMRT dosimetry Freiburg: Conference on Advanced Metrology for Cancer Therapy (CAMCT) [updated Nov 2011; cited 23 March 2014]. Available from: http://www.ptb.de/CAMCT/vortrag/ camct.htm. 6. German Institute of Standards. DIN 6809-8 (Draft) Clinical Dosimetry—Part 8: Dosimetry of small photon fields. Draft. 2014. 7. Das IJ, Ding GX, Ahnesjo¨ A. Small fields: Nonequilibrium radiation dosimetry. Medical Physics. 2008;35(1):206–15.

O053 Calculated spatial response of ionization chambers in air and water D. J. Butler, T. E. Wright, C. P. Oliver Radiotherapy section, ARPANSA, Australia ([email protected]), ([email protected]), ([email protected]) Introduction The spatial response of ionisation chambers to a very small beam of radiation is of interest for modelling the effect of the size of the chamber on measured beam profiles, and to understand the large-field response of the chamber in terms of the response of individual components (such as the walls, stem and central electrode). Method The response of two common chamber types (Farmer and Roos) to a small beam of monoenergetic photons was simulated using Monte Carlo methods. The simulations were performed in 30 cm of air and at 2 cm depth in water. The dose to the air cavity was calculated a function of position of the beam on the chamber. Results Results are shown for the Farmer chamber in air in Fig. 1, for two photon energies.

References 1. Kupfer T, Lehmann J, Franich R, Butler D, Ramanathan G, Rykers K. A large detector for small fields—measurement of dose-area product with a Bragg Peak ionization chamber in stereotactic radiotherapy fields Melbourne: Combined Scientific Meeting (CSM 2014) [cited 15 July 2015]. Available from http://www.csm 2014.com/program/eposters/. 2. Duane S, Graber F, Thomas RAS. Application of dose area product and DAP ratio to dosimetry of IMRT and small field external beam radiotherapy Teddington, UK: National Physical Laboratory

123

Fig. 1 Farmer chamber response as a function of beam position through the centre of the thimble, orthogonal to the thimble axis, without the build-up cap. Two photon energies are shown—50 keV and 110 keV. The outer spikes correspond to the inside of the graphite walls, and the inner spikes correspond to the edge of the aluminium central electrode

Australas Phys Eng Sci Med

O054 Assessment of CT tube current modulation for use in radiotherapy treatment planning J. Hellyer, B. Beeksma, D. P. Truant, S. Arumugam Cancer Therapy Centre, Liverpool and Campbelltown Hospital, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]) Introduction Dose modulation is an automatic exposure control which adjusts the tube current of computed tomography (CT) scans based on the thickness of the patient to maintain high image quality. This technique is well established in diagnostic imaging and results in reductions to patient dose [1, 2]. This study aims to investigate the suitability of dose modulation in radiotherapy where the images are required for treatment planning where consistent CT numbers are required. Method Scan protocols using the Philips Big Bore z-dose modulation setting were compared to current clinical protocols to determine the impact. Functionality was tested using different thicknesses of Solid Water and mAs recorded. The impact of high density materials on current was assessed using wire and ball bearing markers and prosthetic hips attached to different thicknesses of Solid Water. CT number consistency between the standard and dose modulated protocols were tested using a conical wax phantom with inserts of six different materials. Results A reduction in tube current per slice was observed for all dose modulated protocols below a 30 cm diameter (20 % reduction of mAs and dose length product at 30 cm). Above 30 cm, current scaled with thickness. Current also increased for the scans with prosthetics. Patient markers did not influence the dose modulation. The CT numbers for the six materials were consistent between the dose modulated and standard protocols (\2 % variation). Conclusion Dose modulation has minimal impact on CT numbers so can be used for radiotherapy planning scans. Dose modulation reduces the current and therefore patient dose over thin regions and scales upwards to maintain image quality for thicker regions. For bariatric patients this can potentially reduce repeat scans due to poor image quality. References 1. Namasivayam, S. (2006), Optimization of Z-Axis Automatic Exposure Control for Multidetector Row CT Evaluation of Neck and Comparison with Fixed Tube Current Technique for Image Quality and Radiation Dose. AJNR Am J Neuroradiol 27:2221–5. 2. Kalra, M (2005), Chest CT performed with z-axis modulation: scanning protocol and radiation dose. Radiology 237:303–8.

Department of Radiation Oncology, Liverpool and Macarthur Cancer Therapy Centres and Sydney, New South Wales. ([email protected]), ([email protected]), ([email protected]) Introduction In head and neck treatments, the presence of dental fillings may perturb the patient dose distribution. The composition for dental fillings may vary and is often unknown. Scanned CT numbers are limited to high density bone tissue due to the saturation of 12 bit CT numbers at higher densities in most treatment planning systems (TPS). The aim of this work was to investigate how the Tomotherapy TPS accounts for the presence of dental fillings. Method Four high density materials were considered as a dental filling surrogate (lead, stainless steel (SS), titanium and tungsten). Samples of these materials were mounted on an in-house jig with an A1SL ion chamber (Standard Imaging, WI) in a water phantom and CT scanned (Fig. 1). Two Tomotherapy treatment plans were optimised for each high density material separately using a truncated (up to bone density) or an extended (up to lead density) CT density table. Plans with the extended density table both the high density material and surrounding artefact was assigned the appropriate density. Plans with the truncated density table only the artefact was assigned the appropriate density. The dose was measured using the ionchamber and radiochromic films (Gafchromic EBT3) and was compared with the TPS calculated dose for all 8 plans. Results The measured dose for both film and ion chamber agreed within 2 and 5 % when compared with the calculated TPS dose optimised with truncated and extended density tables respectively. The TPS underestimated the dose next to the high density material by 4–5 % and up to 10–30 % when planned with truncated and extended density tables respectively. Conclusion The dose calculated by the Tomotherapy TPS in the presence of surrogate dental fillings showed improved agreement with measured dose when optimized with a truncated density table compared to the extended density table (Figs. 2, 3).

Fig. 1 Experimental setup with transverse CT images. An ion chamber and high density material is mounted on in-house Perspex jig. The two dose level PTV plan was optimised with pseudo RT structures that mimics typical head and neck anatomy

Mean CT Number (HU)

Conclusion The spatial response of a thimble chamber and a planeparallel chamber have been calculated for a range of energies, for 30 cm of air and at 2 cm depth in water. From the response maps, we determine the contribution of the central electrode as a function of energy for the Farmer chamber, and the contribution from areas outside the guard electrode for the Roos chamber.

5000

80 kVp 100 kVp

4000

120 kVp

140 kVp

3000

GSI-14 2000

O055 Accounting for high density dental fillings in head and neck treatment planning using Tomotherapy James Hellyer, Shrikant Deshpande, Philip Vial

1

2

3

4

5

6

ROI Diameter (mm)

Fig. 2 Dependence of mean CT number on ROI diameter and kilovoltage for an aluminium rod of 6 mm in diameter

123

Standard Deviation (HU)

Australas Phys Eng Sci Med 900

80 kVp 600

100 kVp 120 kVp

140 kVp

300

GSI-14

0 1

2

3 4 ROI Diameter (mm)

5

6

Fig. 3 Dependence of standard deviation on ROI diameter and kilovoltage for an aluminium rod of 6 mm in diameter

O056 Study on the Impact of Earthquake on Radiotherapy Centers in Nepal K. Adhikari Objective On 25th April, 2015 devastating earthquake with a magnitude of 7.8 Richter scale and its subsequent aftershocks hit central part of Nepal which killed more than 9000 people and injured around 23,000. The devastating tremor led to the development of fissures in the buildings, houses and other physical infrastructures which also could lead to the possible damage at the radiotherapy bunkers. It is decided to evaluate the radiation level outside and other occupant area around the radiotherapy bunker. This study was done to find out the status of radiation level, shielding structure and the status of radiotherapy equipment. The main objective of this study to find out the impact of an earthquake on Radiotherapy facility. Method The study was done at five different cancer centers in Nepal. This study includes three Linear Accelerators, two Cobalt-60 Teletherapy and four HDR Brachytherapy. Onsite inspection was also done at two new under construction radiotherapy bunkers for Linear Accelerator. Radiation Survey, onsite inspection of the bunker and quality control test was done on Radiotherapy equipment. Result Radiation dose level around Radiotherapy centers shows that all the reference points are within safe limit. The survey result show that all radiotherapy centers were built according to protection criteria. The radiation survey in one center show some dose at the corner of the control console at 20 MV beam but the dose is within the dose limit. At one HDR Brachytherapy bunker, we have found minor hairline crack near to the corner of the bunker. QA test showed that measurement results were within the tolerance level. Keywords Radiation, dose level, Quality Control

O057 ACPSEM Position Paper recommendation for digital mammography quality assurance program— Tomosynthesis addendum P. J. Barnes, L. Cartwright Medical Physics Specialist, MIA Radiology, VIC. ([email protected]), ([email protected]) Introduction Quality control (QC) tests are used to ensure that equipment performs as expected and meets applicable standards. QC testing recommendations for Digital Breast Tomosynthesis (DBT)

123

have not yet been developed for the Australasian environment. This work describes the preliminary investigation of DBT systems, with the aim of providing DBT Dosimetry and Quality Control (QC) recommendations. Method The existing digital mammography tomosynthesis QC recommendations from Europe, America and each DBT system manufacturer were reviewed and evaluated. The evaluation involved performing detailed QC and dosimetry on each DBT system. Results Recommended QC tests for Radiographers include: Artefact evaluation, Image Quality Evaluation and Detector Calibration. Recommended QC tests for annual equipment testing include: Image Quality Evaluation, HVL and MGD, Caliper accuracy and Artefact evaluation. Conclusion An addendum to the existing recommendations1 has been developed to give guidance on the dosimetry, type and frequency of testing for medical physicists, ACPSEM/RANZCR equipment assessors and routine radiography QC. References 1. Heggie JCP et al. ACPSEM position paper: recommendations for a digital mammography quality assurance program v3.0. Australasian College of Physical Scientists & Engineers in Medicine.

O058 Transmitting an effective DICOM real time ultrasound image using modified teleconferencing equipment P. J. Barnes Medical Physics, MIA Radiology, VIC. ([email protected]) Introduction This work presents a method to transmit real time diagnostic quality images using a modified teleconferencing system over a hospital network. Much work has been done previously in telesonography to transmit real-time images without objectively evaluating the quality of the images against well established digital medical imaging criteria. We believe this telesonography project is unique in its approach to transmitting effective DICOM diagnostic quality, real-time ultrasound images. Method Existing and modified BreastScreen methods for evaluating ultrasound images were used to optimise the H264 codec parameters on a specialised monitor. The method included using AAPM TG18 test patterns. ATS Model 551 Small Parts Phantom and ultrasound breast cine clips, simulating transducer movement and breast imaging. Results Evaluations comparing the source ultrasound image and the transmitted ultrasound image were performed. Assessment of the ATS Small Parts Phantom, TG18-LN and TG18-QC2 k test patterns showed good agreement and evaluations met BreastSrceen QC standards. Furthermore, direct comparison of the source breast cine clips and transmitted breast cine clips, comparing anatomical landmarks and pathology were well matched. Conclusion A mammography assessment clinic was conducted remotely from Townsville to Brisbane using the modified teleconferencing system to realise real-time effective DICOM diagnostic ultrasound images. Patient ultrasound images were successfully transmitted over a hospital network using a bandwidth of 900 kbit/s at 30 fps.

Australas Phys Eng Sci Med

O059 Measurement reproducibility for material identification imaging with CT A. Perdomo1,2, Z. Brady1, J. Crosbie2, R. Franich2, J. Du Plessis2 1 Radiology Department, Alfred Health, Melbourne, Victoria. ([email protected]). 2School of Applied Sciences, RMIT University, Melbourne, Victoria. ([email protected]), ([email protected]), ([email protected]), ([email protected])

Introduction The ability to identify a foreign material reproducibly with the use of CT may assist the forensic process. Currently, there are no guidelines on recommended scanning or measurement protocols. Our objective was to assess a range of metal samples with varying densities and sizes to investigate the reproducibility of CT measurements. Method Aluminium, copper and steel rods of various diameters were assessed with both Single Energy (SE) and Dual Energy (DE) CT. A GE Discovery CT750 HD CT scanner (GE Healthcare, Wisconsin, USA) was used to perform an axial scan on each rod individually. Each rod was placed inside a water phantom and imaged at 80, 100, 120 and 140 kVp, with 600 mA and 0.8 s rotation time, as well as using GE’s Gemstone Spectral Imaging (GSI) DE mode. Images were reconstructed using an extended CT scale. Circular regions of interest (ROIs) of varying size were placed in the centre of the rods. The mean CT number (in Hounsfield Units, HU) and standard deviation (SD) were recorded. Results For the materials assessed, the mean CT number and SD of each ROI remains constant for diameters that are less than 50 % of the rod diameter. As the ROI diameter exceeds 50 % of the sample diameter, the mean CT number begins to decrease and SD increases as potential partial volume effects become evident. This is shown in Figures 1 and 2 for a 6 mm diameter aluminium rod. The mean CT number decreases with increasing kVp and the mean CT number for DECT is representative of a scan between 120 and 140 kVp. Conclusion The size of a ROI is important for reproducible measurements of CT numbers for material identification imaging. A ROI diameter of less than 50 % of the object diameter is required for reproducible measurements for the materials studied.

O060 Lead equivalence testing of protective garments Johnny Laban National Centre for Radiation Science, Institute of Environmental Science and Research Ltd. New Zealand. ([email protected]) Introduction Aesthetic issues aside, the two fundamental design considerations for protective garments are weight and protective capability. Unfortunately these characteristics are opposed to each other, and with only weight being immediately apparent to the wearer it is fair to say that the stated protective abilities of garments sometimes fall short of the claims in the interests of producing an attractively light product. Method Although recognised as not being ideal, the protective ability of garments is expressed by the thickness of lead needed to provide the same air kerma attenuation as the garment for a given beam quality. In Australia and New Zealand AS/NZS 4543 (1999) provides a narrow beam measurement methodology for determining lead

equivalence, and requirements for labelling etc. AS/NZS 4543 is in fact a copy of IEC 1331 (1994), which has recently been updated in a 2nd edition to IEC 61331 (2014), and now includes improved measurement methods for assessing lead equivalence such as the use of broad beam geometries, and the specification of X-ray beams more representative of real-world exposure circumstances. Results A range of materials have been tested according to the narrow and broad beam methods of the new IEC 61331. Test results include those for low lead and lead free materials, and the bi-layer materials currently being used. The results of these tests will be presented alongside the stated lead equivalence, with a discussion of some of the more interesting physics issues that arise. Conclusion Actual lead equivalences fall short of the stated claims in a number of instances, and medical physicists involved in purchasing of protective garments should consider the merits of in-house testing to ensure the products they purchase meet local regulatory requirements. References 1. AS/NZS 4543.1-3 (1999): Protective devices against diagnostic medical X-radiation. 2. IEC 1331-1-3 (1994): Protective devices against diagnostic medical X-radiation. 3. IEC 61331-1-3 (2014) Protective devices against diagnostic medical X-radiation.

O061 A portable X-Ray imaging unit for emergency radiography Andrew Edgar, Nicola Winch SCPS, Victoria University, Wellington, New Zealand. ([email protected]) Introduction X-ray radiography remains an essential diagnostic tool for any hospital practice, but is mainly provided through expensive fixed installations operating from the mains power. In some situations, a compact portable, battery powered system would be advantageous. For example, in a major earthquake, the hospital X-ray facilities may be destroyed or out of action for several days at the time when the service is most desperately needed. Such a system would also be valuable in remote locations without a hospital, first responder situations, or in military medicine. Noting the tremendous advances in performance of digital cameras in the past 15 years, we have designed and built a portable system based on optical read-out of storage phosphor and intensifying screens using a semi-professional grade digital camera. Method and Results We have tested CCD and CMOS-based camera systems, but have settled on a CMOS Canon D6 camera for the current version of the system, PXR2.0. The system permits image recording using either CR cassettes or intensifying screens and is built into a hard-shell suitcase. It is completely battery powered, and weighs around 9 kg, providing good portability. The imaging process and read-out takes place through a USB link to a laptop, or by wireless to a smart phone or tablet. Preliminary performance measurements suggest that the 5 Pixel images of size 300 9 220 mm display a spatial resolution of better than 3 lp/mm, and require a dose of 2 times that of typical fixed facilities. The measurements which underpin these and other performance specifications will be presented.

123

Australas Phys Eng Sci Med

The PXR2.0 portable X-ray radiography unit Conclusion A cost-effective, good performance, battery powered, portable X-ray imaging system has been designed and built. References 1. NM Winch and A Edgar, ‘‘X-ray imaging using digital cameras’’, SPIE Medical Imaging (2012), 83135E - 83135E-9.

O062 Difference in using the TRS-398 code of practice and the TG-51 dosimetry protocol for flattening filter free beams J. E. Lye1, D. J. Butler2, C. P. Oliver2 1

Australian Clinical Dosimetry Service (ACDS), Yallambie, Victoria, Australia. ([email protected]). 2Australian Radiation Protection and Nuclear Safety Agency (ARPANSA), Yallambie, Victoria, Australia Introduction The two most commonly used protocols for reference dosimetry in external beam radiotherapy are IAEA TRS-398 and AAPM TG-51. Increasingly flattening filter free (FFF) linacs are in clinical use and published theoretical analysis suggests that a difference of 0.5 % is expected between the protocols (Xiong 2008).

Fig. 1 Graph of the difference in measured output using TRS-398 compared to TG-51

123

Method The Australian Clinical Dosimetry Service (ACDS) has evaluated FFF beam outputs on 11 linacs using both TRS-398 and TG-51 protocols. The response of an NE2561 chamber was modelled using DOSRZnrc. The model studies the difference in kQ in Varian and Elekta linacs when the flattening filter was removed, and when the flattening filter was replaced by a thin metal plate. Results Figure 1 shows the difference from the measured FFF beam outputs using TRS-398 and TG-51 protocols. The modelled FFF kQ as a function of TPR20,10 is 0.6 % lower than the kQ with flattening filter (WFF). This difference is reduced to 0.3 % when considering kQ as a function of %dd(10)x. Thus the measured difference in the TRS-398 and TG-51 protocols should be 0.3 % according to modelling, however the average measured difference is less than 0.1 %. The commercial realisation of FFF beams includes a thin metal filter in the place of the flattening filter. When a 2–3 mm metal plate was included in the model, the difference between the FFF kQ and the WFF kQ was reduced to approximately 0.1 %. Conclusion The average difference between linac outputs measured with TRS-398 and TG-51 protocols was less than 0.1 % for 6 MV FFF and 10 MV FFF. Modelling suggests a 2–3 mm metal plate used in place of the flattening filter offers sufficient filtration for the FFF beam to produce a similar kQ to WFF beams. References Xiong G and Rogers DWO (2008) Relationship between %dd(10)x and stopping-power ratios for flattening filter free accelerators: a Monte Carlo study. Med Phys 35:2104–2109

O063 Comparison of ARPANSA calorimeter with IAEA calorimeter Ramanathan Ganesan, Peter Harty, Jessica Lye, Tracy Wright, Duncan Butler, David Webb Australian Radiation Protection and Nuclear Safety Agency. ([email protected]) Introduction The Australian absorbed dose to water primary standard is a graphite calorimeter which was established as the primary standard for 60Co gamma rays in 1997. In 2009 ARPANSA installed an Elekta Synergy Linear Accelerator (linac) and in 2010 the 60Co source was replaced with an Eldorado 78 treatment head containing a new 60Co source. The ARPANSA calorimeter was re-commissioned on the new 60Co source and on the linac beams. In 2000, the ARPANSA calorimeter developed problems and needed to be repaired. During this time ARPANSA received a similar calorimeter from IAEA on loan as a substitute for the ARPANSA calorimeter. This paper presents the comparison of calorimetric measurements made with both calorimeters at 60Co and linac photon beams. Method Both ARPANSA and IAEA calorimeters were positioned simultaneously with their graphite cores at source-detector distances of 105 cm for 60Co and 110 cm for megavoltage linac photon beams. The calorimetry measurements were done in both quasi-adiabatic and quasi-isothermal modes. The conversion of graphite absorbed dose measured by the calorimeters into water absorbed dose was done by a Monte-Carlo method Lye JE (2013). Results The results of the comparison measurements for dose-tographite Dg and the dose-to-water Dw calculated from Dg are shown in Table below.

Australas Phys Eng Sci Med

TPR20,10

Dg IAEA Calorimeter/Dg ARPANSA Calorimeter

Dw IAEA/Dw ARPANSA

60

Co 6X

0.576 0.673

1.0020 1.0033

1.0022 1.0000

10X

0.734

1.0035

1.0018

18X

0.777

1.0028

0.9993

Conclusion The good agreement of 0.2 % between the calorimeters supports the validity of the ARPANSA calorimeter which is the Australian primary standard of absorbed dose to water. References 1. Lye JE, Butler DJ, Franich RD, Harty PD, Oliver CP, Ramanathan G, Webb DV, and Wright T, Direct MC Conversion of absorbed dose to graphite to absorbed dose to water for 60Co radiation, Rad. Prot. Dos. Vol. 155, No. 1, 100–109, 2013.

O064 Real-time readout of a 3D plastic scintillator array using an inexpensive, shielded CMOS detector M. W. Jennings1,2, T. P. Rutten2, D. Ottaway1 1 School of Physical Sciences, University of Adelaide, Australia. ([email protected]). 2Department of Medical Physics, Royal Adelaide Hospital, Australia. ([email protected]), ([email protected])

Introduction The increasingly dynamic nature of novel radiotherapy techniques indicates the need for real-time assessment of linac dose delivery. This work aims to evaluate the utility of a plastic scintillator array for real-time dosimetry, where the read out is performed with a shielded CMOS camera located inside the treatment bunker. Method The system consists of a cylindrical, water-filled Perspex phantom containing a 3D plastic scintillator array. The phantom is imaged by a low-cost Point Grey Flea3 CMOS camera, shielded using a low melting point alloy enclosure. The camera lens faces away from the phantom and a periscopic mirror arrangement facilitates imaging of the phantom (see Fig. 1). The read, thermal, shot and radiation

Fig. 1 Detector configuration

20000

Mean pixel intensity (ADU)

Beam quality

16000 12000 8000 4000 0 0

50

100

150

200

250

Exposure time (ms) Fig. 2 Linearity of response with exposure time noise sources of the system were analysed and its performance studied. The low-intensity limits of the camera were quantified by varying exposure time beneath a 6 MV, 600 MU/min, 20 9 20 cm2 beam with the central scintillator located at the linac’s SAD. The influence of Cerenkov radiation was also quantified. Results See Fig. 2. The system is able to capture light with an uncertainty of *0.4 % (2r) at a temporal resolution of 500 ms. At 60 ms the uncertainty rises to *5 %, as the read noise of system begins to quench the scintillator signal. Conclusion The uncertainties present in the system due to the various noise sources intrinsic to the system are within acceptable limits. This is aided by the use of radiation shielding and Cerenkov removal techniques.

O065 A study of solid state detectors for kilovoltage X-ray beam dosimetry J. Damodar1, D. Pope2, D. Odgers2, R. Hill2 1 Institute of Medical Physics, School of Physics, University of Sydney, Sydney, Australia. ([email protected]). 2Chris O’Brien Lifehouse, Sydney, Australia. ([email protected]), ([email protected]), ([email protected])

Introduction Kilovoltage X-ray beams have application in therapeutic X-ray units for treating skin cancers, biological irradiators and intraoperative X-ray units. Ionisation chambers are considered the gold standard for dosimetry measurements of kilovoltage X-ray beams[1]. However, they may not be suitable for smaller X-ray field sizes. The purpose of this work is to evaluate a number of small solid state detectors to determine their suitability for kilovoltage X-ray beam dosimetry. Method All dosimetry measurements were performed on a Pantak DXT300 X-ray unit with X-ray beams in the energy range from 50 to 280 kVp. Depth dose measurements were performed with the PTW Advanced Markus parallel plate and Pinpoint ionisation chambers, PTW 60012 diode and PTW 60019 microDiamond detector. The fields sizes used ranged from 2 cm diameter to 12 9 12 cm. Depth doses were also calculated using the EGSnrc Monte Carlo code for the clinical beams. The primary spectra were determined using the SpekCalc program and the geometry of the X-ray tube. Results The benchmark depth doses were determined to be those measured with the Advanced Markus chamber. The agreement

123

Australas Phys Eng Sci Med between the benchmark depth doses and both the PinPoint chamber measurements and the Monte Carlo calculated doses was better than 1–2 %, except at the water phantom surface. This was due to perturbations caused by the presence of the chambers at or slightly above the water surface. For the field sizes studied, the doses measured with the PTW microDiamond were also in good agreement with the benchmark doses to better than 2 % including the 50 kVp X-ray beam. The doses measured with the PTW 60012 diode were in good agreement for the lower energy X-ray beams but gave large dose differences over the first few cm for the 280 kVp X-ray beam. This was attributed to low energy scatter generated by the Compton interactions generated in in the water and measured by diode. Conclusion We have demonstrated that the PTW microDiamond detector and Pinpoint ionisation chambers are suitable for depth dose measurements for kilovoltage X-ray beams. Further investigations are required to evaluate for smaller field sizes where their good spatial resolution are needed. References 1. Hill, R.F., et al., Advances in kilovoltage X-ray beam dosimetry. Physics in Medicine and Biology, 2014. 59(6): p. R183.

O066 Improved interpolative standard for the calibration of thimble chambers for Ir-192 using multiple kilovoltage X-ray beams Ramanathan Ganesan, Viliami Takau, Duncan Butler Australian Radiation Protection and Nuclear Safety Agency, 619, Lower Plenty Road, Yallambie, Victoria 3085. ([email protected]) Introduction HDR Ir-192 sources have wide-spread use in brachytherapy treatments. But, there is no primary standard in Australia for the dosimetry of HDR Ir-192. Traceability in the calibration of HDR Ir-192 may be made by using an interpolated calibration coefficient for a thimble chamber at 250 kV X-rays and at Cs-137 or Co-60 following IAEA Tecdoc 1274 (2002). A more accurate interpolative method is to calibrate the thimble chamber using multiple kilovoltage beams, instead of just the highest energy beam. In this way the response of the chamber to the full Ir192 spectrum can be more accurately taken into account. Details of the establishment of the interpolative standard are presented in this paper. Methods A Farmer chamber (NE 2571) was calibrated for air kerma at various X-ray energies (15–164 keV) and Co-60. The same buildup cap of wall thickness 3.87 mm was used in all calibrations. The air kerma calibration factor for Ir-192 was arrived at by the method recommended by IAEA Tecdoc 1274 by interpolation using the 250 kV and Co-60 calibration factors. Next, published spectra of the three most commonly used HDR Ir-192 sources in Australia/New Zealand viz. Nucletron microselectron-v2(classic), Varian Varisource and Nucletron Flexisource were used for interpolation in which the air kerma rates due to the photon energies from the spectrum were evaluated and divided by the response of the chamber at each energy interpolated from the X-ray calibration coefficients. Results The air kerma calibration coefficients by IAEA Tecdoc method is 41.68 mGy/nC and the spectral interpolation is 41.41 mGy/ nC which is less by *0.7 % for the three sources. This agrees with the results reported by Eduard van Dijk (2004). Conclusion The interpolative standard reduces the uncertainty in brachytherapy dose measurements in the use of HDR Ir-192 sources and provides direct traceability to ARPANSA.

123

References 1. IAEA, ‘‘Calibration of photon and beta ray sources used in brachytherapy,’’ Technical Document 1274 (2002). 2. Eduard van Dijk, Inger-Karine K. Kolkman-Deurloo, and Patricia M. G. Damen, Determination of the reference air kerma rate for Ir 192 brachytherapy sources and the related uncertainty, Medical Physics 31, 2826–2833 (2004)

O067 Energy-resolving performance of a spectral X-ray detector R. K. Panta1, S. T. Bell2, M. F. Walsh2, C. J. Bateman1, R. Aamir1, J. L. Healy3, D. Knight1, N. J. A. de Ruiter1, M. Anjomrouz1, M. Shamshad1, M. Moghiseh1, N. G. Anderson1, A. P. H. Butler1,2,3,4, P. H. Butler2,4,5 1

Department of Radiology, University of Otago, Christchurch, New Zealand. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]). 2 MARS Bioimaging Ltd., Christchurch, New Zealand. ([email protected]), ([email protected]). 3 Department of Biological Sciences, University of Canterbury, Christchurch, New Zealand. ([email protected]). 3 Department of Electrical and Computer Engineering University of Canterbury, Christchurch, New Zealand. 4European Organisation for Nuclear Research, 23 Geneva, Switzerland. ([email protected]), ([email protected]). 5Department of Physics and Astronomy University of Canterbury, Christchurch, New Zealand Introduction Spectral X-ray detectors are used in biomedical imaging to provide high spectral (energy) and spatial fidelity of images. Spectral imaging using X-ray data acquired from multiple X-ray energy ranges allows non-invasive identification and quantification of individual constituents of an object. This work demonstrates the energy-resolving performance of a high-Z semiconductor based spectral detector that allows improving the energy information in biomedical imaging. Methods A CdTe-Medipix3RX1 spectral X-ray detector was operated in an on-chip energy-distortion correction mode (Charge Summing Mode). The energy response of individual pixels of detector was measured using monochromatic c-ray emitted from Am241 and X-ray fluorescence emitted from metallic foils using the technique of Panta et al2. The energy-discrimination performance of the detector was evaluated by performing spectral imaging of a multi-contrast phantom that contained various concentration of iodine, gadolinium and gold based contrast agents, bone-like, and soft tissue-like materials. The phantom was scanned using a MARS scanner 3, 4. Results The energy response function and spectral resolution of pixels (110 x 110 lm2) of CdTe-Medipix3RX detector are well defined when operated in Charge Summing Mode. The K-edge features of various concentrations of iodine (9, 18 and 36 mg/ml), gadolinium (2, 4 and 8 mg/ml) and gold (2, 4 and 8 mg/ml) contrast agents are identified and differentiated. Conclusion The energy-resolving performance of the CdTe-Medipix3RX detector when operated in Charge Summing Mode allows the identification and differentiation of element-specific K-edge features of various concentration of multiple contrast agents simultaneously. This promising result demonstrates the ability to perform elementspecific spectral imaging, which is likely to open up new possibilities

Australas Phys Eng Sci Med of multiple functional imaging by tagging multiple biomarkers with different contrast agents simultaneously.

the patient. A multi-centre international study is underway using this system.

References

References

1. Ballabriga, B. et al. (2013). The Medipix3RX: a high resolution, zero dead-time pixel detector readout chip allowing spectroscopic imaging. Journal of Instrumentation, vol. 8, no. 02, p. C02016. 2. Panta, RK et al. (2015). Energy calibration of the pixels of spectral X-ray detector. IEEE TMI, vol. 34, no. 03, p. 697–706. 3. Anderson, NG et al. (2010). Spectroscopic (multi-energy) CT distinguishes iodine and barium contrast material in mice. European Radiology, vol. 20, no. 09, p. 2126–2134. 4. Walsh, MF et al. (2011). First CT using Medipix3 and the MARSCT-3 spectral scanner. Journal of Instrumentation, vol. 06, no. 01, p. C01095.

1. King, B.W., D. Morf, and P.B. Greer, Development and testing of an improved dosimetry system using a backscatter shielded electronic portal imaging device. Med.Phys., 2012. 39(5): p. 2839–2847. 2. King, B.W. and P.B. Greer, A method for removing arm backscatter from EPID images. Med.Phys., 2013. 40(7). 3. Ansbacher, W., Three-dimensional portal image-based dose reconstruction in a virtual phantom for rapid evaluation of IMRT plans. Med.Phys., 2006. 33(9): p. 3369–3382 4. Fuangrod T, Woodruff H, Van Uytven E, McCurdy BMC, Kuncic Z, O’Connor DJ, Greer PB, A system for EPID based real-time patient dose delivery verification during dynamic IMRT treatment, Med.Phys.40,(9) art. no. 091907, 2013 5. Woodruff HC, Fuangrod T, van Uytven E, McCurdy BMC, van Beek T, Bhatia S, and Greer PB, First experience with real-time EPID based delivery verification during IMRT and VMAT treatments, Int.J. Radiat.Oncol.Biol.Phys. (in press), 2015

IS06 EPID Dosimetry: pre-treatment verification, clinical trial credentialling and real-time patient dose monitoring Peter Greer Calvary Mater Hospital, Newcastle, NSW, Australia. School of Mathematical and Physical Sciences, University of Newcastle, NSW, Australia. ([email protected]) Introduction Electronic portal imaging devices (EPIDs) have been standard equipment on medical linear accelerators for approximately 15 years since flat-panel amorphous silicon imagers became available. Their dosimetric properties are well understood. Some commercial software is available for their use in the clinic mainly for pre-treatment verification and machine QA. In-house methods have also been developed for in vivo dosimetry using EPIDs. In this presentation the methods developed at our clinic for pretreatment IMRT and VMAT QA using EPID will be introduced along with two new applications of EPIDs, remote clinical trial credentialling and realtime patient dose monitoring. Methods An EPID image to dose-in-water plane conversion method was developed using a deconvolution to fluence and in-water dose calculation model1. A correction for EPID arm backscatter is incorporated in this model [2]. An extension of the dose plane to 3D dose distribution allows for calculation of combined field dose in a virtual cylindrical phantom for VMAT deliveries3. The above methods are being applied to remote credentialling of clinical trials in collaboration with TROG Cancer Research for both IMRT and VMAT and Varian and Elekta linacs—VESPA (Virtual Epid Standard Phantom Audit). Remote sites calculate verification plans on the virtual phantoms and deliver the plans to EPID. Both planning doses and EPID images are sent to our institution for EPID conversion to dose and comparison to planning dose. Finally, a system to verify dose delivery in real-time has been developed and implemented (Watchdog) [4–5]. A physicsbased model predicts the transit EPID image frames expected during dynamic deliveries. A frame-grabber system captures the EPID frames and software compares the measured to predicted frames using Chi comparison in real-time. This extends invivo dosimetry post-treatment to during treatment and can detect errors in dose delivery before substantial dose has been delivered to

O068 A dosimetric correction algorithm for intensity modulated radiation therapy pre-treatment verification using an amorphous silicon electronic portal imaging device R. Taylor1, S. L. Elliott2, Jeffrey C. Crosbie3 1 School of Applied Sciences, RMIT University, Australia. ([email protected]). 2WBRC, Alfred Health, Melbourne, Australia. ([email protected]). 3RMIT University, Melbourne, Australia. ([email protected])

Introduction An Intensity Modulated Radiation Therapy (IMRT) pre-treatment verification test is performed by measuring the fluence of individual treatment fields with an amorphous silicon (a-Si) electronic portal imaging device (EPID). A condition of the IMRT plan being classified within tolerance, is subjecting the delivered fluence to a gamma analysis test (3 %, 3 mm) using Varian’s Portal Dosimetry software. Off-axis fields fail this test as a result of the a-Si EPID over responding at extended distances, particularly to low energy photons, facilitating the need of a correction algorithm (Bailey, 2009). Method An open 40 9 30 cm2 field was delivered to the EPID on three Varian (21iX and 21EX) linear accelerators with 6 MV photons. After correcting for pixel sensitivities, the ratio of measured to Portal Dosimetry predicted diagonal dose profiles was determined. A plot of this ratio against the distance from the origin determined the required correction factor. Results Preliminary results show the correction factor for each EPID can be modelled successfully with a polynomial as a function of distance (x), and were calculated to be y1 = 2E - 05x3 - 0.0004x2 + 0.0038x + 0.9997, y2 = 2E - 05x3 - 0.0003x2 + 0.0032x + 0.9995 and y3 = - 8E - 07x4 + 4E - 05x3 - 0.0005x2 + 0.0026x + 0.9999. At a distance of 17.5 cm from the origin, the discrepancies between predicted and measured doses are 6.11, 7.44 and 4.19 % for each EPID respectively. After implementation of the correction factor, discrepancies at this distance are less than 1 %.

123

Australas Phys Eng Sci Med Conclusion Correction factors will be applied to the diagonal dose profiles in the monthly EPID calibration, to correct for off-axis output and improve the portrayal of the delivered fluence for off-axis fields. This is highly beneficial for pre-treatment verification as it will improve the effectiveness of the gamma analysis test by emphasising errors in the planned versus delivered fluence, rather than the EPID itself. The application to clinical plans will be presented. Reference 1. Bailey, DW. et al. 2009, ‘An effective correction algorithm for offaxis portal dosimetry errors’, Medical Physics, vol. 36, no. 9, pp. 4089–4094

O069 EPID back projection for pretreatment QA of complex IMRT K. Croft, T. Groudeva, I. Kostourkov Radiotherapy Department, Palmerston North Hospital. ([email protected]), ([email protected]), ([email protected]) Introduction Verification of IMRT using a small number of beams is easily managed on a field by field basis, however with an increasing number of fields and with rotationally delivered IMRT, looking at QA on a field by field basis becomes impractical. The proportion of patients having complex plans has steadily increased, such that a quick, reliable and easily interpreted method for individual patient QA of treatment plans is needed. Displaying the delivered dose distribution based on EPID results, back projected within a surface rendered model of the patient, allows for the verification results to be presented in a manner readily understandable to the clinicians. Using the EPID as the source for the data enables QA to be performed in a timely manner. Methods Dose planes within the patient model are reconstructed using open field EPID images, corrected for the EPID sag and its characteristic response and compensating for the anatomy via a modified Clarkson algorithm. The accuracy of this IMRT QA method is verified by comparison with dose planes generated by commercial treatment planning systems and with film and ion chamber measurements in a purpose-build water phantom. Results Good agreement between EPID back projection with film, ion chamber and commercial treatment planning prediction has been achieved. Conclusion A time-efficient method for performing patient-specific IMRT QA using EPID images to generate a readily understandable report for the clinicians has been developed and implemented in clinical practice.

O070 Credentialing clinical trials based on EPID dosimetry N. Miri1, J. Lehmann2, B. Zwan3, J. Hatton4, P. Vial5, A. Craig6, V. Beenstock6, T. Molloy7, H. Gustafsson8, P. Greer2 1

School of Physics & Mathematical Science, University of Newcastle, NSW, Australia. ([email protected]). 2Calvary Mater Newcastle, University of Sydney, NSW, Australia. ([email protected]), ([email protected]). 3Gosford Hospital, Gosford, NSW, Australia. ([email protected]). 4TROG Cancer Research, Newcastle, NSW, Australia. ([email protected]). 5 Liverpool Hospital, Sydney, NSW, Australia. ([email protected]). 6Canterbury Christchurch Hospital, Christchurch, New Zealand. ([email protected]), ([email protected]). 7Bloomfield Hospital, Orange, NSW, Australia. ([email protected]). 8Canberra Hospital, Canberra, ACT, Australia. ([email protected]) Introduction This research performs a pilot study applying a remote standard approach for assessment of clinical centres to ensure that they are delivering high quality radiation therapy. The study develops a Virtual EPID Standard Phantom Audit (VESPA) to credential IMRT plans in the centres. Method The method is based on acquiring non-transmission images with an electronic portal imaging device (EPID) by remote centres and data analysis by a central site. CT data sets are provided for the remote site to produce a clinical trial plan using their treatment planning system (TPS). Then, the remote centre transfers the plan to a virtual cylindrical phantom to calculate a reference dose. The plans are then delivered in air and images are recorded with EPID. Finally, the remote centre electronically sends the images and reference dose to the central site. To analyse at the central site, all images of each plan are combined to calculate the 3D dose distribution inside a virtual cylindrical water phantom. An EPID to dose conversion model is used to calculate the dose. The converted dose is then compared to the reference dose in sagittal, coronal and transverse planes using Gamma analysis with criterion of 3 %, 3 mm with a 10 % threshold (Fig. 1). Results The pilot study covered four centres to this date: the data acquisitions of three centres were from Varian linear accelerator with Pinnacle and Eclipse TPS and the other one was from Elekta linear accelerator and Pinnacle TPS. All the centres passed the criterion with the mean pass rates of 98.2, 98.3 and 99.4 % for sagittal, coronal and

References 1. Lorensen W.E. et al. (1987) Marching Cubes: A high resolution 3D surface construction algorithm. Computer Graphics, Vol 21 #4. 2. Kung J.H. et al. (2000) Monitor unit verification calculation in intensity modulated radiotherapy as a dose quality assurance. Med Phys. 27(10) 3. Micke A. et al. (2011) Multichannel film dosimetry with nonuniformity correction. Med Phys. 38(5)

123

Fig. 1 Dose verification for an IMRT plan using the model (left-side) and TPS (right-side)

Australas Phys Eng Sci Med

Mean gamma

Pass rate %

changing gamma criteria was dependent on the resolution of the dosimeter. Pass rates were generally higher for 2 %/3 mm compared to 3 %/2 mm. Conclusion Clinically equivalent action levels determined using linear regression provided a useful guide for clinical decision making and a means to compare findings from different centres where different evaluation criteria are used.

0.32

99.9

References

0.30

99.8

0.31

99.8

95.8

0.39

98.1

98.27

0.33

99.40

Table 1 Multi-centre analysis for dose verification Centres

2D gamma for each plane Sagittal

Coronal

Transverse

Mean gamma

Pass rate %

Mean gamma

Pass rate %

Centre A

0.28

99.7

0.26

Centre B

0.32

99.8

0.27

Centre C

0.42

97.8

0.40

97.6

Centre D

0.50

95.6

0.44

Mean

0.38

98.23

0.34

99.7 100

transverse planes, respectively while the average of mean gamma for each plane was respectively 0.38, 0.34 and 0.33 (Table 1). Conclusion The procedure is practical, fast, remote, consistent, inexpensive and accurate so it promises an efficient and standard method for multi-centre credentialing.

1. Eaton, D. J. (2007) Highly cited papers in Medical Physics. Med Phys 41(8): 080401. 2. Low, D. A. et al. (1998) A technique for the quantitative evaluation of dose distributions. Med Phys 25(5): 656–661. 3. Nelms, B. E. & Simon, J. A. (2007) A survey on IMRT QA analysis. J App Clin Med Phys 8(3): 76–90.

O072 A phantom based study of the effect of intrafraction motion on EPID dosimetry A. L. Fielding1, J. Benitez1, C. E. Jones2 1

Reference 1. King, B. W.et al. (2012).

O071 On the selection of gamma criteria S. B. Crowe1, B. Sutherland2, R. Wilks1, V. Seshadri3, S. Sylvander1, T. Kairn2 1

Royal Brisbane & Women’s Hospital, Brisbane, Australia. ([email protected]), ([email protected]), ([email protected]). 2Genesis Cancer Care Queensland, Gold Coast, Australia. ([email protected]), ([email protected]). 3Princess Alexandra Hospital, Brisbane, Australia. ([email protected]) Introduction The use of the gamma evaluation method for patient specific pre-treatment quality assurance checks of modulated treatments is common—being implemented in most commercial systems. The original gamma evaluation paper (Low et al., 1998) receives the 5th highest number of citations per year of all papers published in Medical Physics (Eaton, 2007). The use of 3 %/3 mm with action levels of 90 % is widespread (Nelms & Simon, 2007), despite recommendations in the literature for the adoption of tighter tolerances. Method This study involved the calculation of gamma agreement indices for 1265 pre-treatment QA measurements using the criteria 1 %/1 mm, 2 %/2 mm, 2 %/3 mm, 3 %/2 mm, 3 %/3 mm and 5 %/ 3 mm. The measurements were performed for intensity-modulated beams, volumetric-modulated arcs and helical tomotherapy treatments from 7 centres, planned using Eclipse, iPlan, Pinnacle and HiArt planning systems, and measured using MapCheck, ArcCheck and EPID dosimetry systems. Results Linear relationships were observed between pass rates calculated using 3 %/3 mm and pass rates calculated using 2 %/2 mm, 2 %/3 mm and 3 %/2 mm, implying that beams failing at 3 %/3 mm would generally fail for the other criteria evaluated—if revised action levels were appropriately selected. The magnitude of the effect of

Science and Engineering Faculty, Queensland University of Technology, Brisbane, Australia. ([email protected]), ([email protected]). 2Radiation Oncology, Princess Alexandra Hospital, Woolloongabba, Australia. ([email protected]) Introduction Electronic Portal Imaging Devices (EPID’s) can be used to verify the dosimetry during the delivery of radiotherapy treatments (1). Previous work by our group has shown how radiological thickness can be used as a surrogate for dose during in vivo EPID dosimetry (2). This work investigates the effect of respiratory motion on the radiological thickness map derived from an EPID dose image. Method Measurements were performed for a 6 MV photon beam using an Elekta Synergy linear accelerator and Elekta iView GT amorphous silicon EPID. The quadratic calibration method was used with images being acquired for plastic water slabs of thickness of 2, 4, 7, 11, 16 and 21 cm. EPID images of a breast treatment field delivered to a CIRS thorax phantom with breast attachment were converted to radiological thickness, using the quadratic calibration. EPID images were also acquired with the phantom on a QUASAR respiratory motion platform for a range of 1 D sinusoidal amplitudes (0.5, 1.0 and 1.5 cm) and frequencies (12, 15 and 20 cycles per minute). EPID images of the phantom using a human respiratory motion trace were also acquired. Monte-Carlo simulations of the linear accelerator, phantom, and EPID were performed to obtain a reference radiological thickness image. Results Measured radiological thickness images for the different amplitudes and frequencies were compared with the thickness image for the static phantom. Analysis of 1D thickness profiles and 2D thickness images showed clear differences for different amplitudes and frequencies of the motion. A clear difference was also seen for the asymmetric human respiratory motion. The viability of using a MonteCarlo simulated reference thickness image was also demonstrated. Conclusion This work has shown it is possible to use the EPID derived radiological thickness for assessing the effect of respiratory motion on the separation for breast radiotherapy treatments. Acknowledgements The authors would like to acknowledge the contributions of Vaughan Moutrie at the Princess Alexandra Hospital, the QUT High Performance Computing and Research Support Group and Elekta for providing technical details for the Monte-Carlo models.

123

Australas Phys Eng Sci Med References 1. van Elmpt W, McDermott L, Nijsten S, Wendling M, Lambin P, Mijnheer B. A literature review of electronic portal imaging for radiotherapy dosimetry. Radiotherapy and Oncology. 2008 Sep;88(3):289–309. 2. Kairn T, Cassidy D, Sandford PM, Fielding AL. Radiotherapy treatment verification using radiological thickness measured with an amorphous silicon electronic portal imaging device: Monte Carlo simulation and experiment. Physics in Medicine and Biology. 2008 Jul;53(14):3903–19.

IS07 Focal radiotherapy using multi-parametric MRI and biological dose optimisation 1,2

A. Haworth

1 Department of Physical Sciences, Peter MacCallum Cancer Centre, VIC, Australia. 2Sir Peter MacCallum Department of Oncology, University of Melbourne, Australia. ([email protected])

Introduction Typical methods to treat prostate cancer aim to deliver ever increasing doses of radiation to the prostate without increasing toxicity. We propose the whole gland approach be replaced with a highly modulated dose distribution with dose planning objectives based on a biological model informed by multi-parametric MRI (mpMRI). Brachytherapy is suggested as an ideal tool for precise dose delivery. Additionally, these methods could be adapted for use with imageguided radiotherapy techniques. Method We have previously demonstrated a biological model can be used to predict treatment failure in patients receiving low dose-rate (LDR) brachytherapy[3]. The model requires knowledge of tumour characteristics such as tumour cell density, proliferation rates and hypoxia. To demonstrate the feasibility of the proposed approach, the methods developed to extract tumour characteristics from mpMRI are described, along with a biologically based inverse optimisation planning process that can achieve optimal prostate tumour control probabilities (TCP) whilst constraining the dose to the urethra and rectum. Results Histology provides ground truth data for deriving the relationship between MRI parametric/pharmokenetic maps and tumour characteristics. Using a novel framework, in vivo mpMRI is co-registered with histology, using a sequence of rigid and non-rigid registration methods. An ensemble-based supervised classification algorithm, trained on textural image features and incorporating cell density estimation has demonstrated a highly sensitive method for automated tumour delineation in histology images [1]. Statistical methods available in R software and MATLAB have demonstrated the potential for the parameter maps to be correlated with tumour characteristics. A 10-patient study demonstrated the biological inverse optimisation algorithm offers significant reductions in urethral doses compared with conventional treatment approaches, with treatment plans demonstrating robustness in the presence of seed displacements [2]. Conclusion Using mpMRI to define tumour location and tumour characteristics opens the way to a truly personalised approach to treating localised prostate cancer. References 1. Weingant M, Reynolds HM, Haworth A, Mitchell C, Williams S DiFranco M. (2015) Ensemble Prostate Tumor Classication in H&E Whole Slide Imaging via Stain Normalization and Cell Density Estimation. MICCAI, in press.

123

2. Betts JM, Mears C, Reynolds HM, Tack G, Leo K, Ebert MA, et al. (2015) Optimised robust treatment plans for prostate cancer focal brachytherapy. Procedia Computer Science, 51. 914–923. 3. Haworth A, Williams S, Reynolds H, Waterhouse D, Duchesne GM, Bucci J, et al. (2013) Validation of a radiobiological model for low-dose-rate prostate boost focal therapy treatment planning. Brachytherapy, 12. 628–36.

O073 On the robustness of IPSA optimised HDR prostate brachytherapy treatment plans J. Poder1, A. Kejda1,2, M. Whitaker1 1 Chris O’Brien Lifehouse, Camperdown, NSW, Australia. ([email protected]), ([email protected]). 2University of Wollongong, Wollongong, NSW, Australia. ([email protected])

Introduction This investigation aims to determine how the dwell time deviation constraint (DTDC) parameter used in the Nucletron Oncentra optimisation algorithm, together with uncertainties in the source dwell time and catheter position, affects the robustness of high dose rate (HDR) prostate brachytherapy plans. Method A total of 10 HDR prostate brachytherapy plans created in Oncentra were re-optimised with a DTDC parameter of 0.0 and 0.4. For each plan, catheter displacements of 3 mm and 7 mm were retrospectively applied and the change in dose volume histogram (DVH) indices for the PTV (V100 %, V150 %, V200 %), Urethra (V110 %, V120 %) and Rectum (V70 %, V80 %) were analysed. The dwell time uncertainty was simulated with the addition of 0.1 s to each dwell position in each plan, and the effect on the PTV indices analysed. Results Changing the DTDC parameter from 0.0 to 0.4 improved the plan robustness for all DVH indices except the PTV V100 %. For a 7 mm displacement, the sensitivities of the individual DVH indexes were dependent on the value of the DTDC. The Urethra V120 % was most sensitive for a DTDC of 0.0 (-6.8 %) and the PTV V100 % for a DTDC of 0.4 (-8.5 %). The Urethra V120 % was the most sensitive to small changes in dwell time for both values of DTDC and there was no significant change in plan robustness when increasing the value of DTDC from 0.0 to 0.4. Conclusion The robustness of HDR prostate brachytherapy plans to uncertainties in catheter displacement and dwell time accuracy in relation to the DTDC was investigated in this study. For uncertainties in catheter displacement, increasing the DTDC improved plan robustness for all DVH indices except the PTV V100 %. However, an increase in DTDC had no significant effect on the plan robustness when taking into account uncertainties in dwell time.

O074 In vivo dosimetry in brachytherapy: Are we ready yet? A. Haworth1,2,3, R. L. Smith3,4, R. D. Franich3 1 Peter MacCallum Cancer Centre, VIC, Australia. 2University of Melbourne, VIC, Australia. 3School of Applied Sciences, RMIT University, Melbourne, Australia. ([email protected]), ([email protected]). 4William Buckland Radiotherapy Centre, The Alfred Hospital, Melbourne, Australia. ([email protected])

Introduction In vivo dosimetry is infrequently used in brachytherapy despite the use of high dose per fraction treatments and the multiple

Australas Phys Eng Sci Med planning and treatment processes that provide an opportunity for treatment errors. Materials and methods A range of detectors have been tested for their suitability for use in brachytherapy applications. Ideally detectors would be small, tissue equivalent, have a high signal to noise ratio, have no energy dependence, provide real time readout, be inexpensive, have a linear dose/dose rate response etc. Whilst a number of detectors have been evaluated for their use in brachytherapy, there currently appears to be no single commercial detector that is capable of providing highly accurate in vivo dose measurement that is suitable for identifying treatment errors across all brachytherapy applications. Results In this presentation we will review the performance of state of the art detectors, potential errors they may catch and how their limitations may be complemented with a comprehensive QA program. We will focus our review on an approach using a flat panel detector which, in a phantom study, has demonstrated differences between measured and planned dose is less than 2 % for 98.0 % of pixels in a two-dimensional plane at an SDD of 100 mm(Smith et al., 2013). Clinical trial results using this system in 6 patients, has demonstrated an ability to capture treatment errors due to a number of process errors. Conclusions Promising non-invasive techniques for high accuracy in vivo dosimetry methods are emerging and may be used in conjunction with a complimentary QA program to offer the possibility of a practical and feasible method of assuring high treatment delivery accuracy and minimising the risk of gross error. Reference Smith, R. L., Taylor, M. L., Mcdermott, L. N., Haworth, A., Millar, J. L. and Franich, R. D. 2013. Source position verification and dosimetry in HDR brachytherapy using an EPID. Med Phys, 40, 111706.

O075 In-vivo spectroscopic dosimetry for a paediatric brachytherapy case

Introduction A 4 year old male patient was admitted to Westmead Children’s Hospital with a diagnosis of prostate rhabdomyosarcoma. Radiotherapy was selected to follow a course of chemotherapy, to avoid mutilating prostatectomy. Due to the age and size of the patient, a temporary, manually afterloaded LDR treatment was developed. The treatment consisted of 16 ProGuide catheters inserted into the prostate, temporarily housing 60 I-125 seeds with activity of 4.4 mCi each for a total prescribed treatment dose of 60 Gy over 4 days. Immediately following insertion, in vivo dosimetry was performed using a spectroscopic dosimeter [1] to confirm the doses to the rectum and several external sites (testicles, hips, buttocks, groin and penis) were not excessive. Method A 15 cm spectroscopic dosimeter with a miniature silicon diode at the tip was used to measure the dose rates along the rectum in 5 mm increments, from the anal verge to a depth of 90 mm. Each measurement point consisted of 3 separate 5 s acquisitions. Measurements at the external sites were obtained by placing the dosimeter tip against the skin, with the same measurement timing employed. Results As shown in Fig. 1, the measured rectal dose profile revealed a maximum dose of 27.5 Gy, compared to a maximum planned dose of 40.3 Gy. Discrepancies between the two profiles were due to limitations in the localisation of the detector within the rectum and slight changes in anatomical positions between CT scan and measurement. External sites received doses between 0.12 Gy (left buttock) and 0.76 Gy (face of penis). Conclusion The measured rectal doses, as well as external pelvic point doses, measured immediately following the insertion of the temporary I-125 seeds met predetermined dose constraints. The novel brachytherapy treatment was hence successfully delivered. Real time dosimeters based on spectroscopic dosimetry will play an important role in immediate quality assurance for unique brachytherapy treatments. References 1. Cutajar DL, Takacs GJ, Lerch MLF, et al., (2006) Intraoperative solid-state based urethral dosimetry in low dose rate prostate brachytherapy, IEEE Transactions on Nuclear Science, Volume 53, Number 3, pp 1408–1412

R. Duncan1, D. L. Cutajar1, K. F. Enari2, A. Howie2, E. Estoesta3, E. Flower3,4, A. Wach3, S. Yau3, G. Busuttil3, J. Poder5, H. Porter2, J. A. Bucci2, J. Kaperlowsky6, V. Ahern3, A. B. Rosenfeld1 1 Centre for Medical Radiation Physics, University of Wollongong, Australia. ([email protected]), ([email protected]), ([email protected]). 2St George Cancer Care Centre, Kogarah, NSW, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]). 3Crown Princess Mary Cancer Centre, Westmead, NSW, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]). 4 Institute of Medical Physics, University of Sydney, Camperdown, NSW Australia. ([email protected]). 5Chris O’Brien Lifehouse, Camperdown, NSW, Australia. ([email protected]). 6 Department of Paediatric Surgery, The Children’s Hospital at Westmead Division of Child and Adolescent Health. ([email protected])

Fig. 1 Measured (black) and planned (red) rectal dose profiles

123

Australas Phys Eng Sci Med

O076 Feasibility of filmless HDR brachytherapy gynaecological applicator commissioning R. L. Smith1, C. Dempsey3,4, M. Hanlon2, J. L. Millar1,2, R. D. Franich1,2 1

William Buckland Radiation Oncology, The Alfred Hospital, Melbourne, VIC, Australia. ([email protected]), ([email protected]). 2School of Applied Sciences, RMIT University, Melbourne, VIC, Australia. ([email protected]), ([email protected]). 3 Department of Radiation Oncology, Calvary Mater Newcastle Hospital, NSW, Australia. 4School of Health Sciences, University of Newcastle, NSW, Australia. ([email protected]) Introduction Commissioning of HDR treatment applicators is essential to ensure accurate reconstruction of the radioactive source path within the applicator. Traditionally this has been undertaken using radiochromic film. The aim of this study was to evaluate the use of a flat panel detector (FPD) to perform treatment applicator commissioning and to highlight additional information that can be determined using this filmless approach. Method Two gynaecological HDR treatment applicators, incorporating interuterine tube/ovoids and an interuterine tube/ring combination were used. Radiochromic film and FPD filmless methodologies were completed for comparison. The applicators were dissembled and secured onto the FPD. A radiograph was acquired of the radio-opaque dwell position markers inserted into each channel. As the HDR source was driven to each dwell position, images of the source radiation were captured by the FPD. These images were processed to determine the position of the radioactive source. Additionally, a radiograph was captured, while the source dwelled, acquiring a double exposure image. A subtraction method was then used (Smith et al., 2013) to visualise the physical source in the applicator channel. Results Both approaches were effective in determining the position of the distal dwell position relative to the external applicator or inner lumen tip. The FPD filmless technique provided an immediate digital result allowing evaluation without moving the applicator. The filmless method also provided the ability to manipulate image sets in order to visualise the actual source within the applicator (Fig. 1). Adjacent dwell positions were also evaluated, which is not easily achievable using a film method. Conclusion This investigation has shown that a filmless treatment applicator commissioning approach using a FPD is feasible and provides additional information not possible with radiochromic film. References

O077 Late rectal bleeding results for prostate cancer patients receiving external beam radiotherapy followed by a HDR brachytherapy boost: Do dose-surface maps provide additional information? C. R. Moulton1, M. House1, V. Lye2, C. Tang2, M. Krawiec2, D. J. Joseph2, J. Denham3, M. A. Ebert1,2 1

School of Physics, University of Western Australia, Crawley, Western Australia. ([email protected]), ([email protected]). 2Radiation Oncology, Sir Charles Gairdner Hospital, Nedlands, Western Australia. ([email protected]). ([email protected]), ([email protected]), ([email protected]), ([email protected]). 3School of Medicine and Population Health, University of Newcastle, New South Wales, Australia. ([email protected]) Introduction This study examines whether dose-surface maps (DSM) provide information that is prognostic of toxicity and not present in dose-volume histograms for prostate cancer patients receiving external beam radiotherapy (EBRT) followed by a high-dose-rate (HDR) brachytherapy boost. Method 167 patients received EBRT in 23 fractions of 2 Gy and HDR in 3 fractions of 6.5 Gy. The EBRT CT was registered to the HDR CT with a rigid + scale + deformable registration in VelocityAI. The EBRT and registered HDR Acuros dose distributions were summed after converting to equieffective doses at 2 Gy/fraction (a/ b = 3). The V1-100 Gy, D1-100 % and D1-10 cc were calculated. Rectum DSMs were obtained by virtually unfolding the rectum surface dose slice-by-slice. Spatial analysis was performed after thresholding the DSMs (1–100 Gy). Patients were classed into toxicity (N = 34) or no toxicity (N = 133) groups if they did or did not have at least a late grade 2 LENT-SOMA rectal bleed. The groups were compared via median comparisons (Mann–Whitney U-tests). Results The V48-90 Gy, D1-29 %, D1 cc, D2 cc and D5 cc were significantly greater for patients with toxicity. Patients with toxicity also had significantly greater widths, lengths and areas for various DSM isodose contours (e.g. 65 and 80 Gy). A number of dose thresholds (e.g. 46 Gy) were significantly more compact for bleeders. Ellipse fits at a number of dose thresholds (e.g. 80 Gy) indicated significantly greater lateral extent, greater longitudinal extent and locations further to the right for bleeders. The proportions of ellipses filled by various thresholded doses (e.g. 46 Gy and 80 Gy) were greater for bleeders.

1. Smith, R. L., Taylor, M. L., McDermott, L. N.,Haworth, A., Millar, J. L. and Franich, R. D. 2013. Source position verification and dosimetry in HDR brachytherapy using an EPID. Med Phys, 40, 111706.

Fig. 1 a Source radiation captured by the FPD. b A double exposure. c The resulting subtracted image showing the physical position of the source in the applicator and path of the drive cable through the applicator lumen

123

Conclusion The parameters and DSMs indicated the doses in highdose regions and the coverage of intermediate and high-dose regions were greater for bleeders. Additionally, the DSMs revealed that doses were located further to the right for bleeders such that doses were more contained within fitted ellipses and more compact.

Australas Phys Eng Sci Med

O078 Comparing the ACDS Level II and III audit outcomes I. M. Williams1, L. Dunn2, J. E. Lye1, J. Lehmann3, J. Kenny4, L. Dunn2, A. C. D. Alves1, T. K. Kron5 1

Australian Clinical Dosimetry Service, Yallambie, Vic, Australia. ([email protected]). 2Olivia Newton-John Cancer and Wellness Centre, Heidelberg, Vic, Australia. 3Calvary Mater Newcastle, Newcastle, NSW, Australia. 4Epworth Health Care, Richmond, Vic, Australia. 5Peter MacCallum Cancer Centre, East Melbourne, Vic, Australia Introduction An audit program which either passes all participants, or conversely, fails all participants provides no information regarding the audited populations and is thus useless. A useful audit populates an image in which the contrast is provided through the variety in audit outcomes that are measured. Extending the image analogy, an audit program is equipped with different tools which act as lenses, capable of focussing on different details, and providing information to the auditors and auditees regarding the causality of the image contrast. The multiple Level ACDS audit design was intended to allow such focussing, and results indicate that it was been successful in doing so. Method The Level II1 and III2 ACDS audits are, respectively, a diagnostic test of the planning system and an end-to-end test of the clinical path, for conformal treatments. As the ACDS audit service has matured the data’s statistical power has improved. The increasing data sets have enabled inter- and intra- audit analysis allowing equipment and algorithm specific correlations to be identified. Results Audit outcome analysis has demonstrated that the Level two audit is particularly sensitive to asymmetric fields, while the Level III audit quantified calculation issues downstream from curved inhomogeneities.

Conclusion The existing ACDS audits are presently offered for conformal treatments. While focussing on the simplest types of calculations, a number of dosimetric issues have been uncovered. The planned extension into IMRT, FFF and rotational therapies References 1. Dunn L., et al., National dosimetric audit network finds discrepancies in AAA lung inhomogeneity corrections. Physica Medica 31, 435–441 (2015). 2. J. Lye, J. Kenny, J. Lehmann, L. Dunn, T. Kron, A. Alves, A. Cole, I. Williams, ‘‘A 2D ion chamber array audit of wedged and asymmetric fields in an inhomogeneous lung phantom,’’ Med Phys

O079 ARPANSA NDRL Service activity and update P. D. Thomas, T. Beveridge, A. Hayton, P. Marks, A. B. Wallace Medical Imaging, Australian Radiation Protection and Nuclear Safety Agency, Yallambie, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected])

Introduction The ARPANSA National Diagnostic Reference Level Service (NDRLS) has been surveying MDCT facilities since August 2011 (Hayton, 2013). An estimated 30 % of CT facilities have registered with the service, logging over 2500 compliant surveys which are used in the development of national DRLs. A survey of administered activities in nuclear medicine was undertaken in 2014/15. A survey of doses in Image Guided Interventional Procedures (IGIP) commenced in 2014. Method Dose data and patient data are collected for a particular imaging protocol on a given imaging device at an imaging facility. Dose metrics are: CTDIvol and DLP (in CT); administered activity (in nuclear medicine); and DAP and dose at the reference point (in IGIP). Patient data include: weight, sex, and age. Additional data describing the imaging protocol are also collected. A facility reference level (FRL) is computed for each such survey as the median value of the dose metric(s). The distribution of FRLs for each imaging protocol is analysed to determine a diagnostic reference level. Results Summary statistics (ARPANSA, 2015) demonstrate a reduction in the 75th percentile of the FRL distributions for MDCT protocols over time. Updated national MDCT DRLs are presently being discussed by a liaison panel and an announcement is expected soon. Work is in progress to enable submission of MDCT data by facilities in New Zealand to the ARPANSA NDRLS. Updated Reference Activities for nuclear medicine are to be announced at the conference of the ANZSNM in March 2016. Survey collection in IGIP is continuing. Conclusion Adoption and monitoring of diagnostic reference levels in Australia is contributing to a reduction in radiation dose in medical imaging. References 1. Hayton, A. et al. (2013) Australian diagnostic reference levels for multi detector computed tomography. Australasian Physical & Engineering Sciences in Medicine. 2013;36(1):19–26. 2. ARPANSA (2015) ARPANSA. National Diagnostic Reference Level Service Statistics Available from: http://www.arpansa.gov. au/services/ndrl/statistics.cfm.

O080 The Australian Clinical Dosimetry Service: Existing findings and the future I. M. Williams1, J. E. Lye1, J. Lehmann2, J. Kenny3, L. Dunn4, A. C. D. Alves1, T. K. Kron5 1 Australian Clinical Dosimetry Service, Yallambie, Vic, Australia. ([email protected]). 2Calvary Mater Newcastle, Newcastle, NSW, Australia. 3Epworth Health Care, Richmond, Vic, Australia. 4Olivia Newton-John Cancer and Wellness Centre, Heidelberg, Vic, Australia. 5Peter MacCallum Cancer Centre, East Melbourne, Vic, Australia

Introduction The Australian Clinical Dosimetry Service, (ACDS) was designed as a pilot program to enable the Australian Government to determine whether a locally designed audit program was suitable for mitigating dosimetric risk to radiotherapy patients within Australia. ACDS findings and recommendations to-date and the future audit plan, incorporating IMRT, FFF, small field and dynamic arc audits will be presented. The presentation will conclude the potential funding models to ensure the ACDS’ future. Method The ACDS was designed over 2009/10 by experts drawn from the three professions in consultation with the national Department of Health. The initial list of requirements over a three year pilot was expressed in a Memorandum of Understanding, (MoU) between

123

Australas Phys Eng Sci Med Health and the Australian Radiation Protection and Nuclear Safety Agency (ARPANSA) which hosted the ACDS. A new 2014 MoU extended ACDS funding and operational targets. Results ACDS audit findings have resulted in numerous facilities modifying both dosimetric practices and internal quality assurance procedures. Additionally, flaws identified in existing planning systems have stimulated the commissioning of replacement planning systems. As the audit program has matured the statistical power of the ACDS data has increased. Higher level audit analysis incorporating 50 audits has enabled the ACDS to identify and quantify algorithm issues, which may be opaque to an individual facility1. These findings, including population analysis for all audit levels will be presented and summarised. Conclusion The ACDS’ future is dependent on stakeholders being convinced that the ACDS provides a risk mitigating and financially viable function within the Australian context. Experience to-date indicates that an audit service functionally mitigates risk for Australian patients, but that the ACDS must continue to develop and optimise its audit service to mitigate the projected risk. An engagement plan for the ACDS’ long term sustainability incorporating ongoing audit development will be presented. References

Fig. 1 Distribution of photon results for the level I audit

1. Dunn L., et al., National dosimetric audit network finds discrepancies in AAA lung inhomogeneity corrections. Physica Medica 31, 435–441, (2015).

O081 Independent verification of Australian dosimetry: The Australian Clinical Dosimetry Service (ACDS) Level I and Level Ib audits A. D. C. Alves, S. Keehan, J. Lye, L. Dunn, J. Lehmann, J. Kenny, M. Shaw, I. Williams Australian Clinical Dosimetry Service, Yallambie, Vic. ([email protected]) Introduction The ACDS remote dosimetric level I audit, using optically stimulated luminescence dosimeters (OSLD), has audited 395 photon beams and 483 electron beams since mid-2012. The service boasts national coverage every two years processing roughly 800 OSLDs per quarter. The more accurate level Ib on-site audit is used to verify dosimetry of new linacs and to verify the OSLDs. Presented are audit findings and issues relevant to on-going audit accuracy. Method Level Ib uses the TRS-398 protocol with Farmer type and Roos chambers with the facility water tank. Level I employs OSLDs in Perspex block phantoms which are posted to facilities for irradiation. OSLDs are read out using the Landauer MicroStar reader (Table 1). Results OSLD data have consistently shown that the average deviation between the ACDS measured dose and the facility reported dose is centred on zero with a standard uncertainty of ± 1.5 %. Recent analysis of the individual batches, nominally 200 OSLDs, has Table 1 Average results for photons and electrons in the level I and level Ib audits along with the standard uncertainty of the measurement Level

Photons

Stand. Unc.

Electrons

Stand. Unc.

I

-0.1 ± 1.5 %

±1.3 %

-0.2 ± 2.0 %

±1.7 %

Ib

-0.2 ± 0.6 %

±0.7 %

0.1 ± 1.0 %

±1.1 %

123

revealed a batch standard uncertainty of ± 1.3 %, and frequent nonzero centroids. Follow up level Ib audits did not support the OSLD findings. 17 OSLDs batches were compared to the ideal normal distribution (Fig. 1). The direct consequence is that the ACDS overreports out-of-tolerance and action level outcomes. Conclusion Intra-batch centroid movement is attributed to instability in the MicroStar reader. A modification to procedure is proposed to make the audit less sensitive to reader instability. Interestingly modelled standard uncertainties were designed to be tighter than expected, to minimise the highest patient risk outcomes which are false negatives.

KS07 The MRI linac in UMC Utrecht, development and current status Bas Raaymakers Department of Radiotherapy, UMC Utrecht, The Netherlands. ([email protected]) The MRI linac, is a 7.2 MV linear accelerator (in short linac) with integrated 1.5 T MRI functionality. In UMC Utrecht, in close collaboration with Elekta and Philips, we designed, developed and prototyped this system. It enables simultaneous MR imaging and radiation delivery. The idea behind the MRI linac developments is to exploit the superior soft-tissue contrast of MRI for real-time image guidance. The initial prototype consisted of a static linac next to the 1.5 T MRI, this set-up was used to demonstrate the proof of concept of simultaneous irradiation and MRI. In the second prototype, the linac was mounted in a ring and equipped with a multi-leaf collimator. The ring is positioned in the mid-transversal plane of the MRI. IMRT and MRI based gating and tracking were demonstrated. The latest phase is the development of a clinical MRI linac system by Elekta, the first pre-clinical prototype has been installed in Utrecht and indeed has also shown to enable simultaneous MRI and irradiation. The clinical introduction of this system will be done in a coordinated fashion via a international consortium. Joint clinical trials in

Australas Phys Eng Sci Med preparation while preceding predicate studies are ongoing for a variety of (prioritized) tumour sites. Also dedicated technical working groups on MRI, adaptive treatment planning, quality assurance and work flow are active to prepare the technical side of the clinical introduction of the MRI linac. An overview of the current status and future plans will be given around the technical status of the MRI linac and the clincal plans of the consortium.

O082 MRI-linac image-guided radiation dosimetry: Initial experiments with the MagicPlate-512 silicon array S. J. Alnaghy1, M. Gargett2, G. Liney3,4, M. Petasecca2, J. Begg3,4, L. Holloway2,3,4,5,6, M. Lerch2, A. B. Rosenfeld2, P. Metcalfe2,3 1

Centre for Medical Radiation Physics (CMRP), University of Wollongong (UOW), Australia. ([email protected]). 2 CMRP, UOW, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]). 3Ingham Institute for Applied Medical Research, Australia. ([email protected]). 4Department of Medical Physics, Liverpool and Macarthur Cancer Therapy Centre, Australia. ([email protected]), ([email protected]). 5Institute of Medical Physics, University of Sydney, Australia. 6South Western Sydney Clinical School (SWSCS), University of New South Wales, Australia. ([email protected]) Introduction MRI-linacs will enable 4D image-guided radiotherapy and require accurate MR visible and compatible dosimeter systems for verification. The MagicPlate-512 (M512); a 2D silicon array detector with 512 sensitive areas (500 9 500 lm2) is optimised for SABR tracking with RF or CT guidance. In this work we discuss device modification required for MRI-linac real-time image-guided dosimetry. Method Motion-tracking utilising silicon detectors has been completed using Calypso RF tracking device (Petasecca, 2015). A similar concept will be employed utilising MR imaging for dynamic detector-tracking (‘dynamic dosimaging’ (Metcalfe, 2014)). The detector was tested for MRI-safety and functionality without irradiation in a 1T fringe field of 3T Siemens Skyra MRI. A tissueequivalent, gel-water phantom (CIRS), providing signal for detector and fiducial visualisation, was utilised to enable MR imaging (fast spin echo sequence). CT images were acquired for electron density. A depth dose comparison in zero magnetic field was completed between our so named gel-water and RMI solid water for 10 9 10 cm2 and 20 9 20 cm2 field sizes (on 6 MV linac). Results The detector functioned at the 1T bore entry position whilst a water phantom was imaged simultaneously at the mid-bore 3T position, with negligible noise seen due to detector RF interference. MR images of a non-powered detector system demonstrated detector visualisation. Detector movements approximating breathing were simulated during dynamic MRI acquisition (fast gradient echo), showing that fiducial markers could be visualised and tracked. For B 10 cm depth, gel-water depth dose data was within 1 % of solid water. Conclusion The current MRI-guided dynamic dosimaging set-up has been demonstrated to be successful in detector visualisation and tracking with a non-powered detector. Future work will consider noise reduction with detector switched on. A MRI-compatible motion platform will be paired with M512. These measurements will be

compared to acquisition in MRI-linac magnetic fields on the MRIlinac device being installed at the Ingham Institute. References 1. Petasecca, M. et al. (2015), Med Phys, 42(6):2992–3004. 2. Metcalfe, P. & Walker, A. (2014), MMND, Port Douglas, Australia. 3. Computerized Imaging Reference Systems Inc (CIRS), Norfolk, VA, USA.

O083 A ground truth for volumetric MRI cardiac tracking using the XCAT phantom N. Lowther1, S. Marsh1, P. Keall2, S. Ipsen3 1

Department of Medical Physics, University of Canterbury, New Zealand. ([email protected]). 2Sydney Medical School, The University of Sydney, Australia. 3University of Luebeck, Germany Introduction A real-time cardiac MRI tracking technique based on template matching has been developed centred on the findings of a cardiac target localisation in real-time MRI investigations [1]. Realtime volumetric tracking of cardiac structures that undergo respiratory and cardiac motion with MRI is required if the novel method of treating atrial fibrillation with radiosurgery is to be realised. MR imaging currently lacks the required spatial and temporal resolution to provide a ground truth for comparison with the real-time cardiac MRI tracking technique. Aim Utilisation of the XCAT digital phantom software [2] to provide a ground truth for cardiac motion to provide verification of the accuracy of the template matching real-time cardiac tracking technique. Method The XCAT digital phantom software allows respiratory and cardiac motion to be simulated, in addition to user defined voxel values for cardiac structures. The left atrium is isolated through thresholding and the volumetric centroid is tracked over a period of multiple respiratory and cardiac cycles. The left atrium 3D centroid position from the XCAT phantom and template matching software are compared. XCAT phantom simulations with varying cardiac and respiratory motion traces are utilised to test the robustness of the realtime cardiac tracking MRI template matching approach. Results Sub-voxel accuracy has been demonstrated in phantoms with small cardiac LR respiratory motion and single voxel accuracy has been demonstrated in phantoms with larger cardiac LR respiratory motion. The presence of a ground truth has allowed optimisation of the template matching approach with dynamic search and absolute motion functions. Conclusion The XCAT digital phantom software has provided verification and identified areas of improvement with the real-time cardiac tracking MRI template matching method. References 1. Ipsen, S., Blanck, O., Oborn, B., Bode, F., Liney, G., Hunold, P.,… & Keall, P. J. (2014). Radiotherapy beyond cancer: Target localization in real-time MRI and treatment planning for cardiac radiosurgery. Medical physics, 41(12), 120702. 2. Segars, W. P., Sturgeon, G., Mendonca, S., Grimes, J., & Tsui, B. M. W. (2010). 4D XCAT phantom for multimodality imaging research. Medical Physics, 37(9), 4902–4915. doi:10.1118/1.348 0985

123

Australas Phys Eng Sci Med

O084 The Australian MRI-Linac program: commissioning of the mega voltage X-ray source J. Begg1, A. George1, S. Alnaghy2, T. Causer2, T. Alharthi3, B. Dong1, G. Goozee1, G. Liney1, L. Holloway1, P. Keall4 1

Department of Medical Physics, Liverpool and Macarthur Cancer Therapy Centre and Ingham Institute for Applied Medical Research, Sydney, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]). 2Centre for Medical Radiation Physics, University of Wollongong, Wollongong, NSW, Australia. ([email protected]), ([email protected]). 3Institute of Medical Physics, School of Physics, University of Sydney, Sydney, NSW, Australia. ([email protected]). 4Sydney Medical School, University of Sydney, Sydney, Australia and Ingham Institute for Applied Medical Research, Liverpool, Australia. ([email protected]) Introduction The Australian MRI-Linac Program[1] aims to develop a 1-T open-bore MRI/6-MV linac. A Varian Linatron MP (Varian Medical Systems, Palo Alto, USA) was selected as the 6-MV X-ray source component of the MRI-Linac system which presents a number of functionality and practical constraints in commissioning compared to a standard clinical linear accelerator. This work aims to commission the Linatron for future experimental and modelling work with and without the external magnetic field. Method Commissioning test were split into safety, mechanical and dosimetry. Safety tests included radiation surveys of the bunker, interlock checks and head leakage measurements. Mechanical checks involved determining the radiation axis of the linatron and aligning lasers and MLCs to the radiation axis. Dosimetry checks included ion chamber characterisation (linearity, reproducibility and proportionality), dose rate dependence on distance, beam quality measurements, profile and percentage depth dose measurements and a preliminary dose calibration. Results The radiation survey indicated environmental dose below legislated limits [2,3]. Audible and visible warnings and interlocks functioned appropriately[4]. Head leakage was higher than IEC limits for medical electron accelerators[5]. In-room lasers and MLCs were aligned to the radiation axis. Linearity was within 1 % of the expected dose above 1-Gy. However, below 1-Gy, a difference of up to 6 % was observed. Reproducibility of 10 9 1-Gy was 0.9 %. A comparison between the sum of 10 9 1-Gy measurements and a 1 9 10-Gy measurement resulted in a difference of 0.2 %. Dose rate decreased following a 1/r2 dependence for distances between 1–4 m. Beam quality was similar to a generic 6 MV linac beam. Conclusion The Linatron has been commissioned for use in on-going MRI-linac development. Characterisation tests established benchmarks in a 0T field and will be repeated inside the magnetic field. Measurements have been used to validate Monte Carlo simulations and determine restrictions on delivery in the MRI-linac system. References 1. Keall, P. J., Barton, M., & Crozier, S. (2014, July). The Australian Magnetic Resonance Imaging–Linac Program. In Seminars in radiation oncology (Vol. 24, No. 3, pp. 203–206). WB Saunders. 2. NSW Government, Radiation Control Regulation, 2003 3. NSW Government, Radiation Control Act, 1990 4. Standards Australia, Guidelines for radiotherapy treatment room designs, 1998

123

5. IEC Pub No: 601-2-1, Safety of medical electrical equipment Part 2: Particular requirements for medical electron accelerators in the range 1 MeV to 50 MeV.

O085 Ongoing development of VMAT patient specific QA methods J. Talbot Cancer and Blood Service, Auckland City Hospital, New Zealand. ([email protected]) Introduction VMAT was introduced for the treatment of prostate cancer at Auckland City Hospital in 2011. Since then it has been used for a variety of larger and more complex treatment sites, which has led to the production of more complex plans. The frequency of VMAT prescriptions has also increased, which has put time pressure on the preparation and measurement processes for QA. There has therefore been an incentive to develop QA techniques to meet the rise in complexity, as well as to reduce workload. Method Gamma comparisons between TPS calculations and cylindrical diode array measurements are performed as part of the QA process. Fields that exceed the dimensions of the array are problematic since the complete cross section of the beam cannot be measured with the array centred on the isocentre. To address this, these plans are delivered multiple times with geometric offsets applied to the array. This is a time consuming process for both measurement and analysis. Ion chamber point dose measurements are also acquired. A plan delivery log-based QA system has been commissioned to alleviate time pressure on staff involved with QA. This automatically assesses plan delivery from linac log files and reduces the time spent on QA preparation, measurement and analysis. Results Large, highly modulated VMAT fields are less likely to produce favourable gamma analysis results. While the treatment log based system has reduced QA workload for a proportion of VMAT plans, many are still assessed dosimetrically with the diode array, including complex plans that are more likely to fail. Ion chamber point dose measurements are the most frequent cause of QA failure, with measurements being systematically lower than calculations. Conclusion VMAT QA methods have improved to meet the increase in demand. Further development of QA techniques is still ongoing.

O086 IMRT fluence complexity calculation T. Hanusova1, V. Vondracek2, K. Badraoui Cuprova2 1

Faculty of Nuclear Sciences and Physical Engineering, Czech Technical University in Prague. ([email protected]). 2 Proton Therapy Center Czech. ([email protected]), ([email protected]) Introduction Fluence complexity of Intensity Modulated Radiation Therapy (IMRT) and Volumetric Modulated Arc Therapy (VMAT) techniques in external beam radiotherapy plays an important role in delivering the right dose to the right place. However, excess fluence complexity causes problems during plan calculation, plan verification and plan delivery. A new method for estimation of fluence complexity in IMRT fields is proposed. Method Unlike other previously published works, it is based on portal images calculated by the Portal Dose Calculation algorithm in Eclipse (version 8.6, Varian Medical Systems) in the plane of the

Australas Phys Eng Sci Med EPID aS500 detector (Varian Medical Systems). Fluence complexity is given by the number and the amplitudes of dose gradients in these matrices. Our method is validated using a set of clinical plans where fluence has been smoothed manually so that each plan has a different level of complexity. Fluence complexity calculated with our method is compared to results of plan verification by means of gamma analysis. Results Fluence complexity calculated with our tool is in accordance with the different levels of smoothing as well as results of gamma analysis, when calculated and measured dose matrices are compared. Conclusion It is possible to estimate plan complexity before carrying out the measurement. If appropriate thresholds are determined which would distinguish between acceptably and overly modulated plans, this might save time in the replanning and remeasuring process. Knowing the level of fluence complexity might help to optimize smoothing methods during planning and improve plan quality. Acknowledgements This work has been supported by a students’ grant no. SGS15/217/OHK4/3T/14 of the Czech Technical University in Prague.

O087 Investigating the impact of treatment delivery uncertainties for advanced lung and head and neck radiotherapy S. J. Blake1, S. A. Arumugam2, T. Ndoro1, L. Holloway2, S. Vinod2, C. Ochoa2, P. Phan2, J. Juresic2, D. Thwaites1 1

Institute of Medical Physics, School of Physics, University of Sydney, Australia. ([email protected]), ([email protected]), ([email protected]). 2 Department of Radiation Oncology, Liverpool & Macarthur Cancer Therapy Centres & the Ingham Institute, Australia. ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]) Introduction The lack of detailed knowledge about treatment delivery uncertainties in advanced RT techniques demands conservative approaches to planning, thus potentially limiting treatment quality. A better understanding of such uncertainties would facilitate further optimisation of advanced techniques with the aim to improve patient outcomes. This study investigates how delivery uncertainties affect plan quality for lung and head and neck (H&N) VMAT plans. Method Reference 6 MV VMAT plans were created for three lung and H&N datasets using the Pinnacle3 TPS on an Elekta Versa linac. In-house code was used to generate a series of modified plans where the reference gantry angle, collimator angle and MLC leaf positions were systematically altered across all control points. Gantry and collimator angles were changed by ± 1, 2 or 5 degrees. Opposing MLC leaves were shifted by ± 1, 2 or 5 mm. Modified plans were read back into Pinnacle and dose calculations were performed. DVH metrics, including the PTV volumes receiving 95 and 100 % of the prescribed dose (V95% and V100% respectively), were extracted from each plan and results for the reference and modified plans were compared. Results For all DVH metrics considered, increasing the magnitude of the plan modification generally resulted in larger deviations from reference plan results. The largest deviations in V95% for the lung and H&N plans occurred for 5 mm MLC leaf shifts and were -16 and -28 %, respectively. Collimator and gantry angle variations typically resulted in deviations to V95% and V100% of \1 %.

Conclusion The impact of treatment delivery uncertainties on VMAT patient dose distributions was investigated with an initial pilot cohort of lung and H&N SABR plans. Target coverage was typically compromised more by changes in MLC leaf positions than gantry or collimator angle. The study is being extended to quantify site- and technique-specific delivery uncertainties. Acknowledgements This work was supported by Cancer Council NSW Project Grant RG 14-11.

O088 Assessment of patient radiation dose in endovascular aortic repair J. Atkinson, B. Khoo Department of Medical Technology & Physics, Sir Charles Gairdner Hospital, WA. ([email protected]), ([email protected]) Introduction Endovascular aortic repairs have improved short term mortality rates for infrarenal and thoracic aneurysms (Bahia et al., 2015). Developments in recent years of endovascular devices have enabled more complex repairs such as for para-visceral and thoracoabdominal aortic aneurysms (Banno et al., 2015). The use of fenestrated and branched endovascular stent grafts offers less invasive repair, but due to the complexity of these repairs, radiation exposure during these procedures can be very high and has the potential to result in skin tissue reactions. Consequently we have reviewed radiation doses from endovascular aortic repairs and investigated contributing factors. Method Retrospective data for 186 patients was extracted for procedures performed on a Siemens Artis Zeego and a Philips FD20 C-arm units. The collected and analysed data included Dose Area Product (DAP), Reference Air Kerma (Ka,r), screening time, procedure type and patient demographic information including weight, height, age and sex. Results DAP values ranged from 29 to 2580 Gy.cm2, with 36 % greater than the 500 Gy.cm2 suggested substantial radiation dose level (SRDL) trigger level for follow-up given by the National Council on Radiation Protection (NCRP, 2010). Fenestrated Endoluminal Vascular Aortic Repair (FEVAR) and Branched Endoluminal Vascular Aortic Repair (BEVAR) procedures resulted in the five procedures with the largest DAP values, which exceed three times 500 Gy.cm2. Very strong correlation was found between DAP and Ka,r, with a notable difference between the two C-arms (r = 0.88 and r = 0.97). As expected, correlation between screening time and both DAP and Ka,r was not strong. Correlation between DAP and weight was not strong (r = 0.39). Conclusion Endovascular aortic repairs vary in complexity and consequently can result in potentially high skin doses. This is particularly true for the complex fenestrated and branched endovascular repairs. Dose optimisation for these procedures and patient follow-up are therefore of great importance. References 1. Bahia SS, et al., Systematic Review and Meta-analysis of Longterm survival After Elective Infrarenal Abdominal Aortic Aneurysm Repair 1969–2011: 5 Year Survival Remains Poor Despite Advances in Medical Care and Treatment Strategies, European Journal of Vascular and Endovascular Surgery (2015), 10.1016/j.ejvs.2015.05.004 2. Banno H, et al., Who should do endovascular repair of complex aortic aneurysms and how should they do them?, The Surgeon (2015), 10.1016/j.surge.2015.03.004

123

Australas Phys Eng Sci Med 3. National Council on Radiation Protection, Outline of Administrative Policies for Quality Assurance and Peer Review of Tissue Reactions Associated with Fluoroscopically-Guided Interventions, NCRP Statement No. 11, December 31, 2014. http://ncrponline. org/Publications/Statements/Statement_11.pdf

O089 Real time operator dose feedback system in interventional cardiology M. K. Badawy1, D. Carrion2, O. Farouque3 1 Department of Medical Sciences, RMIT University, Australia. ([email protected]). 2Independent Scholar. ([email protected]). 3Department of Cardiology, Austin Health, Australia. ([email protected])

Introduction Monitoring staff radiation dose in interventional radiology is a useful way to abide by occupational dose limits. Providing feedback to the staff regarding their dose can positively influence their behaviour and assist them in lowering their occupational dose. The aim of this project is to design and implement a system that provides real-time dose information to the staff, whilst also interacting with the wearer when certain thresholds have been reached. Method A Geiger counter kit using an ionisation chamber is used to form the detector component of the system. This is connected to a Raspberry Pi computer that is worn by the operator at belt level. The computer receives dose readings from the detector and transmits and displays the information on a monitor in the lab. A vibration module is attached to the computer that will vibrate to notify the operator once a dose or time threshold is reached to encourage the implementation of dose saving techniques for the remainder of the procedure. Results A prototype was built and tested on a phantom. The detector is sensitive enough to detect scatter radiation from a phantom both in front and behind ceiling shields and lead skirts. Information collected from the detector was displayed and recorded on a laptop in the angiography suite. This system is aimed for clinical trials towards the end of 2015. Conclusion A successful prototype has been built that can detect and report the dose an operator is receiving during a procedure in real time. The interactive module has the potential to incite implementation of dose saving techniques and lower the occupational dose to staff members in interventional radiology.

O090 Evaluating the influence of radiographers on radiation use in cardiac angiography procedures I. R. Smith1, J. T. Rivers2, R. J. Beevors3, J. Cameron2 1 Physical Sciences, St Andrews War Memorial Hospital, Brisbane, Australia. ([email protected]). 2Queensland Cardiovascular, Brisbane, Australia. ([email protected]), ([email protected]). 3 Cardiovascular Lab, St Andrews War Memorial Hospital, Brisbane, Australia. ([email protected])

Introduction Auditing radiation use in diagnostic coronary angiography (DCA) has usually focused on the role of the cardiologist; however, this model fails to acknowledge the potential influence of the radiographer in these procedures. In this study we evaluate the radiographer’s role in radiation risk management and propose an alternative model for independently assessing their performance.

123

Method Data for a series of 1652 diagnostic coronary angiography procedures performed on one imaging platform by 9 cardiologists and 6 radiographers between January-2012 and June-2015 were analysed using a 2-way analysis of variance (2-way ANOVA) model with cardiologist and radiographer as the independent variables. Metrics evaluated were Fluoroscopy Time (FT), Digital Acquisition (DA) frame count, Dose Area Product (DAP*), Reference Dose (Ref*), Effective Area (DAP/Ref). DAP* and Ref* were adjusted for patient weight. Non-Gaussian distributed data were normalised using a BoxCox transform before analysis. Results Using the 2-way ANOVA model confirmed the differences apparent between the cardiologists when using conventional univariate techniques. When analysis is expanded to assess variations linked to the radiographer, only FT returned a not-significant result (this being the one factor solely controlled by the clinician). For all other radiation measures, there appears to be differences attributable to the radiographers. This analysis did not identify any significant interaction effect. Conclusion The extent to which the radiographer influences procedural radiation use is governed by their level of involvement in each component of the conduct of the imaging procedure. At SAWMH, the radiographer has responsibility for controlling the positing of the C-arm (including angulation as well as focal-skin & patient-detector distance), operation of the contrast injector, DA run lengths, collimation and dose mode selection. This analysis highlights the need to account for, and accommodate the radiographer in programs aimed at monitoring and driving changes in radiation safety in interventional imaging procedures.

KS08 Image guided Radiotherapy D. Jaffray Director, Institute of Health Technology Development, University Health Network (TECHNA), Canada Abstract not yet supplied

O091 Breathing guidance during liver cancer SBRT: impact of audiovisual biofeedback on liver tumour motion Sean Pollock1, Regina Tse2, Darren Martin2, Lisa McLean2, Gwi Cho2, Robin Hill2, Sheila Pickard2, Paul Aston2, Chen-Yu Huang3, Kuldeep Makhija3, Ricky O’Brien3, Paul Keall3 1

Radiation Physics Laboratory, Sydney Medical School, The University of Sydney. 2Department of Radiation Oncology, Chris O’Brien Lifehouse. 3Radiation Physics Laboratory, Sydney Medical School, The University of Sydney Purpose Irregular breathing motion exacerbates uncertainties throughout a course of radiation therapy. Breathing guidance has demonstrated to improve the regularity of breathing motion. This study was the first clinical implementation of the audiovisual biofeedback (AVB) breathing guidance system during liver SBRT. Methods Three liver cancer patients with implanted fiducial markers or surgical clips near the tumour were recruited. Prior to CT sim, patients underwent a screening procedure that decided in situ which breathing condition was best suited for them, either (1) with AVB, or (2) without breathing guidance (free breathing). The regularity of breathing motion was analysed for the screening procedure, CT sim,

Australas Phys Eng Sci Med and each treatment fraction. Tumour motion was obtained from the implanted markers in the CBCT projection images. External motion was obtained from the RPM system. Breathing motion regularity was quantified as the root mean square error (RMSE) in displacement and period. Statistical analysis was performed using unpaired Student’s t test. Results The screening procedure yielded the decision to utilise AVB for two patients; free breathing was chosen for one patient who had naturally regular breathing. For tumour motion, breathing regularity in displacement was improved from 0.22 cm for free breathing to 0.18 cm for AVB (p = 0.23); breathing regularity in period was improved from 0.78 s for free breathing to 0.42 s for AVB (p \ 0.001). For external motion, breathing regularity in displacement was increased from 0.14 cm for free breathing to 0.15 cm for AVB (p = 0.94); breathing regularity in period was improved from 0.68 s for free breathing to 0.46 s for AVB (p = 0.05). Conclusions In the screening procedure, two of the three patients’ breathing regularity in displacement was improved AVB. Across a course of SBRT, AVB also demonstrated to significantly improve the regularity of breathing period over free breathing.

O092 Development of a comparison method for realtime adaptive radiotherapy systems E. Colvill1,2, J. Booth1,3, S. Nill4, M. Fast4, J. Bedford4, U. Oelfke4, M. Nakamura5, P Poulsen4, E Worm4, R Hansen4, T Ravkilde6, J Scherman Rydho¨g7,8, T Pommer9,10, P Munck Af Rosenscho¨ld7,8, S Lang11, M Guckenberger11, C Groh12, C Herrmann12, D Verellen13, K Poels13, L Wang14, M Hadsell14, T Sothmann15, O Blanck16,17, P Keall1 1

University of Sydney, Sydney, Australia ([email protected]). 2Royal North Shore Hospital, Sydney, Australia ([email protected]). 3North Shore Hospital, Sydney, Australia ([email protected]). 4The Institute of Cancer Research and The Royal Marsden NHS Foundation Trust, London, UK ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]). 5Kyoto University, Kyoto, Japan ([email protected]). 6Aarhus University Hospital, Aarhus, Denmark ([email protected]). 7Rigshospitalet, Copenhagen, Denmark ([email protected]). 8Niels Bohr Institute, University of Copenhagen, Copenhagen, Denmark ([email protected]). 9Rigshospitalet, Copenhagen, Denmark. 10Karolinska University Hospital, Stockholm, Sweden ([email protected]). 11University Hospital Zurich, Zurich, Switzerland ([email protected]), ([email protected]). 12Wu¨rzburg University, Wu¨rzburg, Germany ([email protected]), ([email protected]). 13Universitair Ziekenhuis Brussel, Vrije Universiteit Brussel, Brussels, Belgium ([email protected]), ([email protected]). 14Stanford University, Palo Alto, USA ([email protected]), ([email protected]). 15University Clinic Eppendorf, Hamburg, Germany ([email protected]). 16 University Clinic Schleswig–Holstein, Kiel, Germany. 17Saphir Radiosurgery Center, Gu¨strow and Frankfurt am Main, Germany ([email protected]) Introduction Real-time adaptive radiotherapy is becoming more wide spread with patients treated via CyberKnife (since 2004), Vero (2011) and MLC tracking (2013) technology, with couch tracking planned to be clinical in 2015. To date no comparisons methods have been created to encompass all real-time radiotherapy techniques. In

this study we have developed a common set of tools and methods for benchmarking real-time adaptive radiotherapy. Method Ten institutions with CyberKnife, Vero, MLC or couch tracking technology were involved in the study. The common materials were anonymized lung and prostate CT and structure sets, patient-measured motion traces (four lung, four prostate) and SBRT planning protocols (lung: RTOG1021, prostate: RTOG0938). The institutions planned and delivered lung and prostate plans to a moving dosimeter programmed with tumour motion. For each trace the plan was delivered twice; with and without motion adaptation, each measurement was compared to the static dosimeter dose and the percentage of failed points for c-tests recorded. Results Ten institutions developed and implemented a benchmarking method, including two CyberKnife, two Vero, four MLC and two couch tracking sites. For all lung traces all sets show improved dose accuracy from a mean 3 %/3 mm c-failrate of 0.4 % with correction and 6.2 % with no motion correction (p \ 0.001). For all prostate traces the mean 3 %/3 mm c-failrate was 0.4 % with correction and 5.9 % with no motion correction (p \ 0.001). The difference between the four adaptive systems was small with an average 3 %/3 mm cfailrate of \1 % for all systems with adaption for lung and prostate. Conclusion A common set of tools has been developed for comparing real-time adaptive radiotherapy systems and a multi-institutional study performed. The results show the systems all account for realistic tumour motion accurately and performed to a similar high standard, with real-time adaptation significantly outperforming non-adaptive methods.

O093 Feasibility of scatter imaging guidance during lung cancer stereotactic ablative radiotherapy (SABR) James C. H. Chu1, Gage Redler1 1 Rush University Medical Center, Chicago, IL 60612, USA ([email protected]), ([email protected])

Introduction Early stage lung cancers can be effectively treated by stereotactic ablative radiotherapy (SABR). The effectiveness of this treatment, however, depends on the ability to deliver large radiation doses to the target volume with millimeter accuracy. We present a feasibility study of a novel imaging modality for this application. Method Compton scatter is a natural by-product of external beam radiation therapy. The distribution of scattered radiation contains patient anatomy information and, when measured from multiple directions, can be used to reconstruct the patient treatment geometry in real-time. We have developed an analytical model to describe a pinhole camera-based scatter imaging system. Monte Carlo N-Particle (MCNP) code was used to simulate phantom images produced by scatter from 6 MV treatment beams for different imaging geometries. Experimental scatter images were measured with a Konica CR flatpanel detector for various phantoms on a Varian TrueBeam linear accelerator. Phantoms used include cylindrical phantoms of water, lung, and bone densities and a clinical lung tumor phantom containing chest wall, lung, and tumor equivalent materials. Results The amount of scatter measured varied with scattering angle, in agreement with both analytical model calculations and Monte Carlo simulation results. Measured image density from three cylindrical phantoms also closely matched those predicted theoretically. Lung tumor phantom images showed excellent signal and contrast. For example, at 90  scattering angle, the signal-to-noise-ratio (SNR) for the tumor ROI was 8.9 and 149.9, respectively, for 10 MU and 5000 MU exposures (corresponding to 0.5 and 250 s exposure time).

123

Australas Phys Eng Sci Med The contrast-to-noise-ratio (CNR) was 4.1 and 29.1, respectively, for these same ROIs. Conclusion This study has demonstrated the feasibility of using scatter imaging to provide image-guided SABR treatments to early stage lung cancer patients. Scatter imaging presents potential benefits of real-time high quality image guidance without requiring additional radiation.

O094 Dosimetric effects to static lymph nodes target for high-risk prostate cancer patients underwent multileaf collimator tracking radiotherapy treatment 1,2

3

2,4

3,4

Y. Ge , A. Bashiri , E. Colvill , J. Booth , P. Keall

2

1

Kolling Institute, Northern Sydney Local Health District, St Leonards NSW 2065 Australia. 2Radiation Physics Laboratory, Sydney Medical School, University of Sydney, Australia ([email protected]), ([email protected]). 3 Institute of Medical Physics, School of Physics, University of Sydney, Australia ([email protected]), ([email protected]). 4Northern Sydney Cancer Centre, Royal North Shore Hospital, Sydney, Australia ([email protected]) Introduction Real-time multi-leaf collimator (MLC) tracking for intrafraction prostate motion has been implemented in a world-first 28-patient clinical trial (Keall et al. 2014). For high-risk patients with static pelvic lymph nodes to be treated simultaneously with the moving primary prostate tumour, the nodal coverage could be compromised due to the relative motion of the two targets while the whole beam is shifted during MLC-tracking (Fig. 1). This study estimated the dosimetric effects to the static structures for the high-risk patients in the MLC-tracking trial. Method Eighteen fractions with large beam displacements during MLC tracking treatment of four high-risk trial patients were selected for this study. The effect of setup correction for interfraction prostate motion was estimated by recalculating the dose with isocenter shifted 4 and 8 mm in anterior, posterior, superior and inferior directions. The effect of intrafraction MLC-tracking was estimated using the dose reconstruction method (Poulsen et al. 2012) that has been developed and validated for estimating delivered dose in dynamic treatment. The actually delivered doses to the static structures, including nodal clinical target volume, small bowel and femoral head were estimated and compared to the original planned doses.

Fig. 1 For high-risk prostate patients with involved static nodes and moving prostate tumour, the MLC-tracking treatment (c) ensures the primary prostate tumour coverage but may deliver deviated dose to the nodal target when compared with the original plan (a) or the standard care without intrafraction adaptation (b). The outline indicates the clinical target volumes and the filled area indicates the delivered radiation dose

123

Results Compared with the original plan, the maximum nodal D99 % changes for 4 and 8 mm interfraction motion are 4 and 20 % decrease, respectively, both in superior direction. The maximum nodal D99 % change due to the intrafraction motion is 5.3 % decrease for a patient with prostate motion mainly in anterior–superior direction. Two fractions exceeded the dose limit (V45Gy \ 100 cc) for small bowel. Conclusion This study shows that the doses to static anatomies for patients underwent MLC-tracking need to be evaluated separately. The effects of the treatment adaptation depend on the direction and magnitude of the beam adaptation and the variance is not clinically significant for the selected fractions. References 1. Keall, P. J., E. Colvill, R. O’Brien, J. A. Ng, P. R. Poulsen, T. Eade, A. Kneebone and J. T. Booth (2014). ‘‘The first clinical implementation of electromagnetic transponder-guided MLC tracking.’’ Medical Physics 41(2): 020702. 2. Poulsen, P. R., M. L. Schmidt, P. Keall, E. S. Worm, W. Fledelius and L. Hoffmann (2012). ‘‘A method of dose reconstruction for moving targets compatible with dynamic treatments.’’ Medical Physics 39(10): 6237–6246.

O095 Time-resolved dosimetry using a PinPoint ionisation chamber for patient-specific VMAT quality control measurements Robert J. W. Louwe1, Thomas W. S. Satherley1, Rebecca A. Day1, Lynne Greig1 1

Wellington Blood and Cancer Centre, New Zealand ([email protected]), ([email protected]), ([email protected]), ([email protected]) Introduction The causes of deviations observed during patientspecific QC of VMAT are usually difficult to uncover because only the integrated fraction dose is measured. This hampers efficient management of out-of-tolerance results. Therefore, we have implemented a method where the dose delivered between the individual control points is measured. This allows correlation of observed deviations to planning and delivery parameters, and enables the cause of deviations to be found. Method Patient-specific QC was performed for 95 VMAT plans for various treatment indications using a TrueBeam (Varian Medical Systems, Palo Alto, USA). The VMAT plans were generated using Varian’s TPS, Eclipse v10–v11. Measurements were performed using PinPoint detectors (PTW types 31014, 31015) in a plastic water slab phantom (CIRS, Norfolk, USA) at positions representative of the PTV and OARs. Data acquisition and analysis was performed using inhouse developed software. Results The overall average deviation of the measured integral fraction dose relative to the computed TPS dose was -0.3 ± 2.7 % (1 S.D.). Plotting deviations per control point as a function of the distance to the field edge (DTFE) showed that the largest deviations were observed when the detector was located near a field edge. These deviations depended on the detector position relative to specific sides of the MLC leaves and could be related to limitations of the TPS beam model. Additionally, analysis of the results highlighted both sub-optimal modelling of the treatment couch and measurement errors.

Australas Phys Eng Sci Med

0.06%

0.06%

0.06%

0.03%

0.03%

0.03%

0.00%

0.00%

(a)

-0.03%

-3

-2

0.00%

(b)

-0.03%

-0.06% -1

0

1

2

-0.06% 3 -3

delivery for all complexity levels was well within clinical acceptance levels. Results of all eight cases will be presented at the conference. Conclusion Within this study, VMAT H&N plan quality could be improved without degrading the dose calculation and delivery accuracy.

-2

(c)

-0.03%

-1

0

1

2

-0.06% 3 -3

-2

-1

0

1

2

3

References

Average contribution to the deviation per fraction as a function of the DTFE [cm] for detector positions: (a) at the MLC leaf tip, (b) at MLC leaf sides facing, or (c) not facing the focal spot of the Linac. Blue solid lines represent the deviations measured with EBT film dosimetry at the corresponding field edges. Conclusion Use of time-resolved patient-specific QC can uncover the causes of observed deviations, and provide information that can be used to improve the accuracy of VMAT.

1. Varian Medical Systems, Palo Alto CA, USA. 2. Sun Nuclear Corporation, Melbourne FL, USA. 3. R.J.W. Louwe et al., Time-resolved dosimetry using a pinpoint ionization chamber as quality assurance for IMRT and VMAT. Med Phys. 2015; 42(4):1625–39.

O096 The impact of plan complexity on the accuracy of VMAT for the treatment of head and neck cancer

So-Yeon Park1, Sung-Joon Ye2, Hong-Gyun Wu1, Jong Min Park1

Thomas W. S. Satherley1, Rebecca A. Day1, Robert J. W. Louwe1 1

Wellington Blood and Cancer Centre, New Zealand ([email protected]), ([email protected]), ([email protected]) Introduction The complexity of VMAT H&N plans in our department is presently limited by a MU constraint to ensure the accuracy of patient treatment within the range of the departmental experience. Loosening the MU constraint, and setting more stringent optimisation objectives on OARs and PTVs could potentially yield higher quality treatment plans but may also degrade the accuracy of the TPS calculation or the plan delivery at the treatment machine. The aim of this study is to investigate the relation between plan complexity, plan quality, and the accuracy of TPS calculation and delivery. Method Clinical plans of eight previously treated H&N patients were selected as the baseline lowest complexity level (C1). Approximate pareto-optimal C3 plans were created focusing on sparing spinal cord, brain stem and parotid glands while maintaining target coverage. Additionally, C2 plans were created with plan quality in between C1 and C3. All plans were developed using Eclipse v11(1). The average leaf pair opening (LPO), critical leaf pair opening (%LPO \ 1 cm) and mean leaf travel were used as plan complexity metrics. Plan verification measurements using a Varian TrueBeam(1) included ArcCheck(2), time-resolved point dose measurements(3) and film dosimetry. Results Initial results for five cases (Table 1) showed that increasing plan complexity from C1 to C3 reduced the Spinal Cord Dmax, Brain Stem Dmax and Parotid Gland Dmean up to 12.9 Gy, 11.1 Gy and 7.4 Gy, respectively. In addition, the target coverage (D98 %) of the target volumes could be increased up to approximately 1.0 Gy. Verification measurements showed that the plan calculation and Table 1 TRPD, EBT film and ArcCheck results; c-analysis was performed using a {2 %; 2 mm} criterion Complexity level

Mean point dose Mean EBT film c Mean ArcCheck deviation ±1 SD pass-rate ±1 SD c pass-rate ±1 SD

C1

-0.1 ± 1.0 %

95 ± 3 %

96 ± 4 %

C2

-0.1 ± 1.2 %

95 ± 2 %

95 ± 3 %

C3

0.0 ± 1.2 %

96 ± 2 %

94 ± 5 %

O097 Texture analysis on the edge-enhanced fluence of VMAT

1 Department of Radiation Oncology, Seoul National University Hospital, Seoul, Republic of Korea ([email protected]), ([email protected]), ([email protected]). 2Program in Biomedical Radiation Sciences, Department of Transdisciplinary Studies, Seoul National University Graduate School of Convergence Science and Technology, Seoul, Republic of Korea ([email protected])

Introduction Texture features of edge-enhanced fluence were analysed to quantify modulation degree of volumetric modulated arc therapy (VMAT) plans. Method Twenty prostate and 20 head and neck VMAT plans were retrospectively selected. Fluences of VMAT plans were generated by integration of monitor units shaped by multi-leaf collimators (MLCs) at every control point. When generating fluences, the values of pixels

Fig. 1 Edge-enhanced and non-enhanced fluence of VMAT. The fluences with non-edge-enhancement of prostate (a) and head and neck (H&N) volumetric modulated arc therapy (VMAT) plans (c) are shown. Those fluences were generated by whole integration of every monitor units (MUs) shaped by multi-leaf collimator (MLC) apertures at each control point (CP). The fluences with edge-enhancement of prostate (b) and H&N VMAT plans (d) are also shown. For edgeenhancement of fluences, when integrating MUs, the values of pixels (size of 1 mm 9 1 mm) representing MLC tips were doubled

123

Australas Phys Eng Sci Med representing MLC tips were doubled to prevent smearing out of small or irregular fields (edge-enhancement) described in Fig. 1. Six kinds of textural features which were angular second moment, inverse difference moment, contrast, variance, correlation and entropy were calculated with particular displacement distances (d) of 1, 5 and 10. Plan delivery accuracy was evaluated by both global and local gamma-index methods, mechanical parameter differences and differences in dose-volumetric parameters with linac log files. Spearman’s rank correlation coefficients (rs) were calculated between the values of textural features and VMAT delivery accuracy. Results The rs values of contrast (d = 1) with edge-enhancement to global gamma passing rates with 2 %/2 mm, 1 %/2 mm and 2 %/ 1 mm were 0.546 (p \ 0.001), 0.744 (p \ 0.001) and 0.487 (p = 0.001), respectively. Those with local 2 %/2 mm, 1 %/2 mm and 2 %/1 mm were 0.588, 0.640 and 0.644, respectively (all with p \ 0.001). The rs values of contrast (d = 1) to MLC and gantry angle errors were -0.853 and 0.655, respectively (all with p \ 0.001). Contrast (d = 1) showed statistically significant rs values in 11 dose-volumetric parameter differences from a total of 35 cases. It showed generally better correlations to plan delivery accuracy than previously suggested textural features with non-edge-enhanced fluences as well as conventional modulation indices. Conclusion Contrast (d = 1) with edge-enhanced fluences could be used as modulation index for VMAT. References 1. Otto K: Volumetric modulated arc therapy: IMRT in a single gantry arc. Medical Physics 2008, 35310–317. 2. Park JM, Park SY, Kim H, Kim JH, Carlson J, Ye SJ: Modulation indices for volumetric modulated arc therapy. Physics in Medicine and Biology 2014, 59 7315–7340. 3. Webb S: Use of a quantitative index of beam modulation to characterize dose conformality: illustration by a comparison of full beamlet IMRT, few-segment IMRT (fsIMRT) and conformal unmodulated radiotherapy. Physics in Medicine and Biology 2003, 48 2051–2062. 4. McNiven AL, Sharpe MB, Purdie TG: A new metric for assessing IMRT modulation complexity and plan deliverability. Medical physics 2010, 37505–515. 5. Li R, Xing L: An adaptive planning strategy for station parameter optimized radiation therapy (SPORT): Segmentally boosted VMAT. Medical physics 2013, 40 050701. 6. Scalco E, Fiorino C, Cattaneo GM, Sanguineti G, Rizzo G: Texture analysis for the assessment of structural changes in parotid glands induced by radiotherapy. Radiotherapy and Oncology : Journal of the European Society for Therapeutic Radiology and Oncology 2013, 109 384–387. 7. Park SY, Kim IH, Ye SJ, Carlson J, Park JM: Texture analysis on the fluence map to evaluate the degree of modulation for volumetric modulated arc therapy. Medical physics 2014, 41 111718.

IS08 The role of diagnostic radiology patient dosimetry in diagnostic reference levels I. D. McLean1 1

Medical Physics and Radiation Engineering, The Canberra Hospital, Australia ([email protected]) Introduction Recent epidemiology publications have increased public awareness about radiation doses from diagnostic radiology. Dose reduction, through technique optimisation, utilises diagnostic reference levels (DRLs) derived from dose indicators, typically

123

associated with patient measurements over a sampled population. Patient dosimetry on the other hand, can result in an individual dose quantity, of known accuracy, from which a risk estimate can be made. A principle difference between this and a current DRL dose indicator is that the former should properly include the consideration of patient size. Method Of five specific applications for diagnostic radiology dosimetry [1], perhaps fluoroscopy and computed tomography (CT) are of the most importance to clinical medical physicists, while the population of paediatric patients is also of immediate concern. Standard CT dosimetry can be augmented with the use of size specific dose estimation factors [2] to convert phantom dose indicators to more realistic estimates of effective dose. This process, along with others [3], can be applied to paediatric populations, where effective dose estimation is most critical, and difficult, exacerbated the relatively low numbers of childhood CT examinations (approximately 2 % of total). Results The effect of size has a critical impact on both DRL dose indicators and effective dose. Many current CT scanners do collect patient size indicators which in the future may impact on DRL determination. Size indicators are currently not readily available on fluoroscopic equipment where skin dose estimation and particularly patient dose from cone beam procedures may be compromised. Conclusion The effect of size cannot be ignored for effective patient dose estimation. This parameter may be incorporated into the use of DRLs in the future, particularly for paediatric populations. References 1. INTERNATIONAL ATOMIC ENERGY AGENCY, Dosimetry in Diagnostic Radiology: An International Code of Practice, edn. 457, T., IAEA Rep. TRS 457, Vienna (2007). http://www-pub.iaea.org/ MTCD/publications/PDF/TRS457_web.pdf. 2. AMERICAN ASSOCIATION OF PHYSICISTS IN MEDICINE, Size-Specific Dose Estimates (SSDE) in Paediatric and Adult Body CT Examinations, AAPM Rep. 204, New York (2011). http://www.aapm.org/pubs/reports/RPT_204.pdf. 3. INTERNATIONAL ATOMIC ENERGY AGENCY, Dosimetry in Diagnostic Radiology for Paediatric Patients, Human Health Series No. 24, IAEA Rep. Human Health Series No. 24, Vienna (2013). http://wwwpub.iaea.org/MTCD/Publications/PDF/Pub1609_web. pdf.

IS09 Medical Imaging: low level radiation and risk M. L. Bartlett1, John D. Mathews2, Anna V. Forsythe2, Darren Wraith3, Zoe Brady4 1

Dept of Nuclear Medicine, Royal Brisbane & Women’s Hospital, Australia ([email protected]). 2School of Population and Global Health, University of Melbourne, Carlton, Vic 3053, Australia. 3Institute of Health and Biomedical Innovation, Queensland University of Technology, Kelvin Grove, 4059, QLD, Australia. 4Radiology, Alfred Health, Prahran, Vic, 3181 Australia As the number of medical imaging procedures continues to grow, there is increasing interest in quantifying the risk associated with low levels of ionising radiation. Arriving at a robust risk estimate is difficult for such low doses, but recent studies have reviewed large cohorts over several years, and a picture is now emerging. In the UK, a follow-up study of 180,000 young people exposed to CTs between 1985 and 2002 found the risk of leukaemia and brain cancer increased with dose1. In Australia, children and adolescents who received Medicare services were tracked between 1985 and 2005. Of these,

Australas Phys Eng Sci Med 680,000 received CTs. Cancer incidence was 24 % higher in the exposed than the unexposed group2. CT scans have a characteristic pattern of high dose rate, arising only from photons, and usually to a restricted body part. In contrast, most low level radiation exposure is low dose rate, whole body exposure and may involve a range of radiation types. This pattern is more characteristic of nuclear medicine studies, and the risk profile from nuclear medicine is now under investigation using the same cohort of Australian children and adolescents who had Medicare scans between 1985 and 2005. For each nuclear medicine scan identified from Medicare, the dose was estimated based on the radiopharmaceutical, the administered activity, and the age of the patient. Outcomes were determined by linkage to the Australian Cancer Database and the National Death Index. Results indicate an increased incidence or risk of cancer for those exposed compared to those unexposed after allowing for a lag period between exposure and diagnosis and further adjustments for age, sex and year of birth. The risk also appears to increase for each additional nuclear medicine procedure. Comparison of risks between CT and nuclear medicine exposures is still being undertaken. Reference 1. Pearce et al. Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retrospective cohort study. Lancet 2012;380:499–505. 2. Mathews et al. Cancer risk in 680,000 people exposed to computed tomography scans in childhood or adolescence: data linkage study of 11 million Australians. BMJ 2013;346:f2360.

O098 Investigation of integral dose for Kilovoltage Intrafraction Monitoring (KIM)-guided MLC Tracking for Lung SABR T. Shaw1, P. Juneja1,2, V. Caillet1,2, J. Martland2, E. Colvill2,3, P.J. Keall1, J.T. Booth1,2 1 School of Medicine or Physics, University of Sydney, Australia ([email protected]). 2Northern Sydney Cancer Centre, Sydney, Australia. 3School of Medicine or Physics, University of Sydney, Australia.

Introduction The objective of this study is to assess the additional kV imaging dose from KIM against the potentially reduced treatment dose from 4D planning methods applied with MLC tracking. We compared lung SABR dose to KIM-guided MLC tracking SABR integral-dose. Method The kV settings for a 125 kV x-ray beam from Varian OBI were investigated allowing visual inspection of the fiducials in CIRS thorax phantom in the presence of MV scatter. The kV dose at 15 9 15 cm field-size was measured with a 0.6 cc thimble chamber using the phantom according to IPEMB protocol. Measurements were performed in ipsilateral-lung, contralateral-lung, heart, and spine for right-sided and left-sided lesions. These were used to retrospectively estimate overall imaging dose for the whole treatment of 9 lung SABR patients. For these patients, treatment plans for standard SABR and simulated MLC tracking were used to create ‘typical’ plans on the phantom. The treatment dose for MLC tracking delivery was summed (linearly) with kV dose from KIM to give the tracking integral-dose and this was compared against the standard SABR dose.

Results The kV setting of 0.2 mAs and 5 fps at 125 kV provided clear visual inspection of the lung fiducials. This beam was characterised with 4.59 mm Al HVL. KIM dose (cGy) ranges for the whole treatments were: 5–16 (ipsilateral-lung), 0–9 (contralateral-lung), 0–13 (heart) and 0–6 (spine). Integral doses were on average reduced by 278 cGy in ipsilateral lung and 32 cGy in contralateral lung with KIM-guided MLC tracking. Similarly, the heart and spine integral doses were reduced by, respectively 89 and 60 cGy. Conclusion KIM-guided MLC tracking lung SABR enabled a net reduction in doses to normal tissue compared to standard delivery and thereby can contribute to safer application of the KIM system. Additional patient plans and film dosimetry is currently being investigated to strengthen these findings.

O099 The first kilovoltage intrafraction monitoring trial for gated prostate radiotherapy: Geometric and dosimetric results J. T. Booth1,2, P. Juneja1,2, V. Caillet1, J. A. Ng2, E. Colvill1,2, R. T. O’Brien2, T. N. Eade1, A. B. Kneebone1, P. R. Poulsen3, P. J. Keall2 Northern Sydney Cancer Centre, Sydney, Australia. 2School of Medicine or Physics, University of Sydney, Australia ([email protected]). 3Aarhus University Hospital, Denmark

1

Introduction Kilovoltage intrafraction monitoring (KIM) utilises a single kV imager during radiotherapy delivery to determine a realtime prostate trajectory. KIM is currently being investigated in a phase I prospective clinical trial of gated prostate VMAT. The goal of this study is to investigate the geometric and dosimetric impact of gated prostate cancer radiotherapy based on KIM. Method Seven patients (out of planned 30) have been recruited. Simultaneous intra-fraction kV and MV images from the 242 fractions were used offline to measure the accuracy of the KIM real-time measurements. The dose delivered during each fraction with KIM was calculated using an isocentre shift dose reconstruction method. For the fractions where a gating event (with motion exceeding 3 mm-for5 s) occurred, dose calculation was also performed for a simulated treatment scenario with no KIM gating correction. Dosimetric impact of KIM gating was evaluated through comparison of target and normal tissue dose volume statistics. Results In total, 35 out of 255 fractions required KIM gating corrections. Prostate motion (mean ± standard deviation) with KIM corrections and simulated with no KIM corrections were 2.0 ± 1.1 mm and 3.7 ± 1.4 mm respectively. The mean error ± standard deviation of KIM in the LR, SI and AP directions was 0.1 ± 0.5, 0.1 ± 0.4, and -0.3 ± 0.5 mm respectively. Mean (range) differences between the planned and KIM corrected doses versus the planned and no KIM correction doses were respectively: PTV D95 % -1.2 (-3.4, 0.7) & -2.3 (-10.9, 0.4); CTV D100 % -1.4 (-4.4, 0.7) vs. -0.7 (-7.0, 1.9); rectum V65 % -2.7 (-9.4, 5.7) vs. -4.3 (-11.2,18.7); and bladder V65 % 1.8 (-1.7, 8.0) vs. 3.1 (-2.9, 13.3). Conclusion The clinical implementation of real-time KIM image guidance combined with gating for prostate cancer improves both the geometric and dosimetric agreement between the planned and delivered treatments. Accuracy and precision of the KIM is submillimetre.

123

Australas Phys Eng Sci Med

O100 Potential dosimetric benefits of MLC tracking for lung SABR V. Caillet1, E. Colvill1,2, K. Szymura1, M. Stevens1, J. Booth1,2, P. Keall1 1

Northern Sydney Cancer Centre, Sydney, Australia ([email protected]). 2School of Medicine or Physics, University of Sydney, Australia Purpose The objective of this study was to investigate the dosimetric benefits of multi-leaf collimator (MLC) tracking for lung SABR treatments compared to a conventional ITV-based approach. Method The 10 most recent single lesion lung SABR patients were re-planned following a real-time adaptive protocol. The protocol consists of a 4D-PTV defined as the GTV delineated at end-of-exhalation plus 5.0 mm margins. Plans were optimized adhering to the standard dosimetric constraints of the conventional SABR lung protocol. Treatments were prescribed so that 48-50 Gy was delivered to C98 % of the PTV in 4–5 bi/tri-weekly fractions. For each patient, the mean doses to the OARs for the conventional ITV and real-time adaptive plans were quantified and statistically compared. The dosimetric accuracy, as assessed by GTV coverage, for both plans was evaluated using an isocenter dose reconstruction method (Colvill 2015, Poulsen 2012). The Calypso transponders positioned into a box phantom was mounted onto the HexaMotion programmable motion device (ScandiDos, Uppsala) and tracked with the Varian Calypso system and in-house tracking software to collect the MLC positions. The dynalogs were used as input for dose reconstruction to calculate the treated dose and the dose that would have been delivered had MLC tracking not been applied. Results Reductions in mean dose to the OARs were observed for the real-time adaptive plans relative to the conventional SABR plans with an average (range) reduction for the ipsilateral (deducting the GTV) lung of 28 % (10–57 %), contralateral lung 29 % (6–83 %), spinal cord 34 % (-17–169 %) and heart 47 % (5–167 %). Significant differences (p \ 0.05) were found between the lung doses of the conventional SABR and real-time adaptive plans. Conclusion Application of MLC tracking for lung SABR patients has the potential to significantly reduce the dose to OAR during radiation therapy while maintaining target coverage. Results from the delivered dose distributions are currently under analysis. References 1. Colvill, Emma, et al. ‘‘MLC Tracking Improves Dose Delivery for Prostate Cancer Biology* Physics (2015).Radiotherapy: Results of the First Clinical Trial.’’ International Journal of Radiation Oncology*. 2. Poulsen, Per Rugaard, et al. ‘‘A method of dose reconstruction for moving targets compatible with dynamic treatments.’’ Medical Physics 39.10 (2012):6237–6246.

([email protected]). 3Aarhus University Hospital, Denmark ([email protected]) Introduction The first MLC tracking clinical treatments have been performed for a prospective prostate clinical trial. The aim of this study is to assess whether MLC tracking improves the agreement between the planned and delivered doses for prostate cancer radiotherapy compared to standard no tracking treatment delivery. Method MLC tracking has been implemented in a 28 patient prostate cancer radiotherapy trial [1], with 858 treatment fractions delivered. The prostate trajectory and treatment MLC positions were collected during each fraction and the data used as input for dose reconstruction (isocenter shift method) [2,3]. The treated dose (with MLC tracking) and the modelled no tracking dose were both reconstructed. The percentage difference from planned for PTV, CTV, rectum and bladder dose-volume points were calculated for each fraction. Analysis of variance using the F-test was used to test whether MLC tracking improves the consistency between the planned and treated MLC tracking values over that of no MLC tracking. Results The results for the first fifteen patients showed that the average mean prostate motion during beam-on for 275 fractions was 1.4 mm, with 95 % of fractions having a mean displacement of \2.5 mm. The mean difference and standard deviation between the planned and treated MLC tracking doses, and the planned and no tracking doses were: PTV D99 % -0.8 ± 1.1 % vs. -2.1 ± 2.7 %; CTV D99 % -0.6 ± 0.8 % vs. -0.6 ± 1.1 %; rectum V65 % 1.6 ± 7.9 % vs. -1.2 ± 18 %; and bladder V65 % 0.5 ± 4.4 % vs. -0.0 ± 9.2 % (p \ 0.001 for all dose-volume results). Conclusion The results of this study show that improved consistency between the planned and delivered doses is obtain when implementing MLC tracking compared to doses that would have been delivered without MLC tracking. References 1. Keall PJ, Colvill E, O’Brien R, Ng JA, Poulsen PR, Eade T, Kneebone A, Booth JT. The first clinical implementation of electromagnetic transponder-guided mlc tracking. Medical Physics 2014;41:020702. 2. Poulsen PR, Schmidt ML, Keall P, Worm ES, Fledelius W, Hoffmann L. A method of dose reconstruction for moving targets compatible with dynamic treatments. Medical Physics 2012;39:6237–6246. 3. Colvill E, Poulsen PR, Booth JT, O’Brien RT, Ng JA, Keall PJ. Dmlc tracking and gating can improve dose coverage for prostate vmat. Medical Physics 2014;41:091705.

O102 Respiratory gating based on internal electromagnetic motion monitoring during liver SBRT P. R. Poulsen1, E. S. Worm1, R. Hansen1, L. P. Larsen1, C. Grau1, M. Høyer1

O101 Dosimetric results of the first MLC tracking prostate trial E. Colvill1,2, J. Booth1,2, R. O’Brien1, T. Eade2, A. Kneebone2, P. Poulsen3, P. Keall1 1 University of Sydney, Sydney, NSW, Australia ([email protected]) ([email protected]). 2Royal North Shore Hospital, Sydney Australia ([email protected]), ([email protected]), ([email protected]),

123

1

Aarhus University Hospital, Aarhus, Denmark ([email protected])

Introduction Intra-fraction motion may compromise the target dose in SBRT treatments of tumours in the liver. Respiratory gating can improve the treatment delivery, but gating based on external surrogates may be inaccurate. In April 2015, we treated our two first liver SBRT patients with Calypso-guided respiratory gating at Aarhus University Hospital. Method Two patients with solitary liver metastases were treated in three fractions with respiratory gated SBRT guided by three implanted electromagnetic transponders. The treatment was delivered

Australas Phys Eng Sci Med in end-exhale with beam-on when the centroid of the three transponders deviated less than 3 mm (LR and AP directions) and 4 mm (CC) from the planned position. Log files were used to determine the transponder motion during beam-on in the actual gated treatments and in simulated non-gated treatments. The motion was used to reconstruct the delivered CTV dose distribution with and without gating. The reduction in D95 (minimum dose to 95 % of the CTV) relative to the plan was calculated for both treatment courses. Results With gating the maximum course mean (standard deviation) geometrical error in any direction was 1.2 mm (1.8 mm). Without gating the course mean error would mainly increase for Patient 1 (to -2.8 mm (1.6 mm) (LR), 7.1 mm (5.8 mm) (CC), -2.6 mm (2.8) (AP)) due to large systematic cranial baseline drifts at each fraction. The errors without gating increased only slightly for Patient 2. The reduction in CTV D95 was 0.5 % (gating) and 12.1 % (non-gating) for Patient 1 and 0.3 % (gating) and 1.7 % (non-gating) for Patient 2. The mean duty cycle was 55 %. Conclusion Respiratory gating based on internal electromagnetic motion monitoring was performed for two liver SBRT patients. The gating added robustness to the dose delivery and ensured a high CTV dose even in the presence of large intra-fraction motion. Data will be presented for all treated patients (four patients at time of writing).

Overall per plan QA time reduced from approximately 60–20 min. Plans are assessed on M3D-TPS agreement for mean PTV dose, PTV 90 % dose coverage and 3D gamma analysis. The ability to visualise DVHs, isodoses and gamma analysis in patient geometry is a paradigm shift for plan QA. This clinical relevance is clearly advantageous, with the important caveat that differences arising from systematic algorithm differences should be distinguished from calculation ‘‘errors’’. It must be recognised that these differences were not previously identified by point dose calculations or most common QA measurements. A calculation system clearly cannot assess for delivery issues, however calculation algorithm and MLC modelling differences typically manifest as dose differences when there is excessive complexity, which can be a useful trigger for deeper scrutiny. Conversely, in at least one instance an IMRT plan failed QA by diode array measurement but passed M3D assessment. Conclusion Software based patient specific QA presents both advantages and challenges. A deliverability check should still be incorporated and overall system QA should be reviewed. Software does not replace robust commissioning and periodic measurement, but can add value to routine patient specific QA with a significant efficiency benefit.

O104 ‘‘FluDo’’ : A fluence to dose checker for IMRT QA B. J. Cooper1, S. Geoghegan1 1 Canberra Hospital, Canberra, Australia ([email protected]), ([email protected])

O103 Transitioning away from measurement based patient specific QA – the epworth radiation oncology experience J. Kenny1, D. Jolly1, A. Perkins1, S. Atkins1, E. Jhala S. Towns1 1

Epworth Radiation Oncology, Epworth Healthcare, Melbourne, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]) Introduction For complex treatments, a patient specific measurement to verify dose calculation accuracy and deliverability is the standard means of performing a treatment planning system (TPS) independent check. This is a significant rate limiting step in the patient care path and for a mature program, the commonly used methods have significant time and resource implications but only add questionable value to overall treatment plan quality. Method Since March 2015, Epworth Radiation Oncology has phased in a software based system using Mobius 3D (M3D) and FractionLab (Mobius Medical Systems, Houston) for 3D dose calculation and dynalog analysis, respectively. Results Over 300 patient and phantom plans have been assessed with M3D to date.

Introduction Patient specific IMRT QA often requires machine access and medical physics labour. Detector array instruments (e.g. MapCheck) have been the mainstay for IMRT fluence verification. We propose an in-house software verification tool which utilises Varian ‘‘dynalog’’ files to reconstruct delivered fluences into planar dose for comparison against Pinnacle planar dose calculations. We present ‘‘FluDo’’: a fluence-to-dose software tool, potentially reducing medical physics labour for routine IMRT QA. Method Pylinac is a Python software package which performs several tasks including reconstructing opening densities of IMRT beam delivery from Varian dynalog files. Gaussian kernels were convolved with the output fluence reconstructed from the dynalog file of an IMRT beam delivery using the Pylinac ‘‘log_analyzer’’ module creating increasingly ‘‘blurry’’ versions of the delivered IMRT beam using 6 MV photons. The standard deviation (r) term for the Gaussian kernel was varied over the range 2.0 to 6.0 mm in steps of 0.1 mm. Each of the reconstructed ‘‘blurry’’ delivery maps is compared against the planar dose computation of the IMRT beam from Pinnacle (100 cm SAD at 5 cm depth in water), and the difference metric value plotted against the standard deviation (r). Results Our in-house software, ‘‘FluDo’’ takes dynalog files from delivered IMRT beams and successfully produces virtual delivered dose maps. A Gaussian standard deviation of r = 3.1 mm gives the minimum difference between the ‘‘delivered’’ and TPS calculated IMRT beams (Fig. 1). A ‘‘FluDo’’ output dose difference map is shown (Fig. 2).

123

Australas Phys Eng Sci Med

Fig. 1 A page of the electronic check sheet created for a test patient plan. The check sheet is filled out by two physicists, each with their own set of checkboxes Conclusion ‘‘FluDo’’ has been demonstrated to create virtual delivered dose maps though programming and connecting open software tools with the potential to reduce medical physics labour for IMRT QA. References 1. Pylinac https://github.com/jrkerns/pylinac/tree/master/pylinac

O105 An Electronic Checklist for Physics QA using Eclipse Scripting and C# K. Roozen1, P. Ramachandran1 1 Peter MacCallum Cancer Centre, Vic, Australia ([email protected]), ([email protected])

Introduction The AAPM task group 53 report1 recommends that as a modern treatment planning system (TPS) is complex, measurement of data for quality assurance (QA) is needed. As such, a series of treatment check-sheets have been developed for pre-treatment plan QA. The checklists are physical documents that cannot be easily copied to guard against the risk of loss and over time a large collection of documents results that is difficult to store and indexed. By developing an electronic check-sheet that results in a soft copy, there are benefits such as data backup and electronic analysis, and potential

123

efficiency gains in data gathering that can be time consuming and restricted by access to terminals. Method Using the Varian Eclipse planning system (version 11.0) scripting application programming interface2,3 (ESAPI), raw plan data was retrieved directly from the TPS database without having to source it from the application. The raw data was then loaded into a windows form application (Visual Studio C# 2010) which formatted the data into digitally reproduction of the physics QA check-sheet. Results An electronic check sheet filled using the plan data gathered for a test patient is shown in Fig. 1. The check sheet can be filled from a stored file or directing from the Eclipse database. The assigned physicist can edit and save the checklist to be later reviewed by a supervisor, who can add further comments, approve, and create PDF report. Conclusion The electronic check sheet is currently being used for IMRT physics QA for prostate and head and neck cancer IMRT. It is being adapted for stereotactic ablative body radiotherapy (SABR) and electron Monte Carlo calculation (eMC) checks. It is a comprehensive plan checker that allows streamlining of physics QA. References 1. AAPM task group 53: Quality assurance for clinical radiotherapy treatment planning 2. Eclipse scripting API reference guide (2011), downloaded from my.varian.com. 3. Varian developer’s forum (https://variandeveloper.codeplex.com).

Australas Phys Eng Sci Med

O106 Commissioning of Mobius3D, a novel independent verification software D. Jolly1, A. Perkins1, J. Kenny1 1 Epworth Radiation Oncology, Epworth Healthcare, Melbourne, Australia ([email protected]), ([email protected]), ([email protected])

Introduction Mobius3D (M3D) is a novel software package for independent verification of radiation therapy treatment plan calculations. It is an automated solution that utilises patient DICOM data to calculate full 3D dose using a collapsed cone algorithm. It presents metrics comparing M3D dose to that calculated by the treatment planning system, including mean target doses, DVH data and voxelby-voxel gamma comparisons. This is not only more comprehensive than traditional MU check software but also fundamentally different, and thus new and more comprehensive commissioning processes are required, this process will be presented herein. Method Commissioning started with simple beam data comparisons between local reference data and that presented by M3D, including PDDs, OARs and OFs for all photon energies and a wide range of field sizes and depths. Simple phantom calculations for various combinations of field size, MLC, EDW and off axis fields were compared, followed by a broad range of 3DCRT clinical patient plans. IMRT plans were compared with ion chamber measurements and adjustments to the MLC modelling made where appropriate. Results Good agreement in beam data was observed with the maximum difference being 1.7 % and the average difference 0.1 %. Phantom calculations showed good MU and mean volumetric agreement, although some gamma disagreements were observed in the build-up and penumbral regions, especially for 10 MV photons. The 3DCRT patient calculations had a mean PTV difference of 0.8 and 3 %, 3 mm gamma pass rate of 95.3 %. Slight adjustments were required for the MLC modelling for all photon beams, and following this good agreement was also observed. Conclusion M3D has been successfully commissioned and released for clinical use as the primary method of independent plan verification at Epworth Radiation Oncology. Investigations are ongoing to commission M3D for specialised techniques such as dynamic conformal arcs, SRS cones and brachytherapy.

O107 Early clinical experience and case studies with Mobius3D, a novel independent verification software D. Jolly1, A. Perkins1, S. Atkins1, E. Jhala1, S. Towns1, J. Kenny1 1 Epworth Radiation Oncology, Epworth Healthcare, Melbourne, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected])

Introduction Mobius3D (M3D) is a novel software package for independent verification of radiation therapy treatment plan calculations. It is an automated solution that utilises the full patient DICOM data (plan, CT, structure set and dose) to independently calculate 3D dose using a collapsed cone algorithm. It presents metrics comparing M3D dose to that calculated by the treatment planning system, including mean target doses, DVH data and voxel-by-voxel gamma comparisons. M3D has been used as the primary independent verification software for 3DCRT and IMRT patients at Epworth Radiation

Oncology (ERO) since March of 2015. Clinical experience with such a platform will be presented herein. Method As an independent verification tool, M3D provides much more data than traditional point based software tools, and thus uncertainties associated with patient dose calculations are more apparent than ever before. A summary of all patient data will be presented, with specific case studies for multiple brain metastases, hippocampal sparing, breast, SABR lung and the ACDS thorax audit phantom. Results The ongoing clinical use of M3D has shown dose calculation differences in regions where there has traditionally been a higher level of uncertainty, namely in the build-up, out of field and inhomogeneous media such and lung and bone. Because M3D highlights the effects of dose differences on the target and organs at risk, the clinical relevance of such differences can be assessed. Due to the software’s automation and clinical relevance the feedback from daily users (physicists, radiation therapists and oncologists) has been positive. Conclusion M3D has been successfully implemented into the clinical workflow as the primary method of independent plan verification at ERO. Having volumetric, clinically relevant dosimetric data has allowed clinicians to make informed decisions regarding the potential uncertainties associated with patient dose calculations.

O108 Comparison of paediatric CT doses for common examinations in Queensland, Australia S. H. Aliuddin1, D. L. Thiele1, G. McGill1, M. Irvine1 1 Biomedical Technology Services, Royal Brisbane & Women’s Hospital, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected])

Introduction Recent studies and campaigns have highlighted the importance of optimisation of paediatric CT doses with regards to minimising the risk of future cancer induction. However many public medical imaging facilities are challenged to monitor paediatric CT doses due to low patient numbers. Method Three paediatric anthropomorphic CT phantoms (representing 1yo, 5yo and 10yo) were imaged on 27 public diagnostic CT scanners in Queensland according to local age-based protocols for head, chest and abdo-pelvis studies. Radiation dose metrics (e.g. CTDIvol and DLP) were recorded for each study. Results were separately analysed based on the reconstruction type (i.e. filtered back projection or iterative) available on the CT scanner. With regards to comparison to national DRLs the worst case was 40 % of scanners exceeding the child head DRL when imaging the 10yo phantom. Use of iterative reconstruction did not necessarily result in lower doses. Conclusion Overall the study has shown wide ranging radiation doses indicating poor optimisation of scanners for a paediatric population. However the outliers are easily identified for further investigation. Typical ranges have also been determined that will assist in developing local achievable dose levels. References 1. Image Gently Campaign retrieved from The Alliance for Radiation Safety in Pediatric Imaging, http://www.imagegently.org 2. The 2011–2013 National Diagnostic Reference Level Service Report (2015), A Wallace et al., TRS 171, ARPANSA Feb 2015

123

Australas Phys Eng Sci Med 3. Mathew J et al. (2013) Cancer risk in 680,000 people exposed to computed tomography scans in childhood or adolescence: data linkage study of 11 million Australians, BMJ 346:f2360 4. Hayton A et al. (2013) Australian per caput dose from Diagnostic Imaging and Nuclear Medicine, Radiation Protection Dosimetry 156:445–450 5. Pearce MS et al. (2012) Radiation Exposure from CT scans in childhood and subsequent leukaemia and brain tumours; a retrospective cohort study, Lancet 380:499–505 Acknowledgments Children’s Health Queensland Royal Melbourne Children’s Hospital Study, Education and Research Committee (SERC) Queensland

O109 Monte Carlo calculated organ dose reduction from CT with tube current modulation using WILLIAM, a paediatric voxel model M. Caon1 1 School of Health Sciences, Flinders University, Adelaide, Australia ([email protected])

Introduction It is unusual for the whole body of a child to be imaged with CT for diagnostic purposes. Consequently there are few complete CT data sets of children that span the anatomy of a single individual from head to toe. A source of whole body medical image data sets are the low resolution CT scans performed by combined PET-CT scanners. Most published Monte Carlo calculated organ doses from CT do not simulate helical scans or include tube current modulation (TCM) in their calculations. This talk describes a paediatric voxel model (WILLIAM) constructed from low resolution CT images and its use to calculate relative dose reduction when helical CT scanning is simulated with TCM implemented compared to without. Method 368 low resolution CT images of a 7 year old male patient were segmented to produce a voxel model of anatomy from head to feet. The EGS4 Monte Carlo code was used to simulate radiation transport and energy deposition through WILLIAM for a CT chest and for a CAP CT exam both with TCM implemented and without TCM implementation. TCM was simulated in both the angular (x–y plane) and axial (along the z axis) senses. Results For the chest CT exam, the thyroid, lungs, adrenal glands, oesophagus, thymus and heart were entirely within the beam and the relative dose to these organs when TCM was implemented decreased by between 53 and 60 % compared to the doses without TCM. For those organs fully within the beam for the CAP exam, the reduction in relative doses ranged from 43 % for the testes and gall bladder to 62 % for the heart. Conclusion Low resolution CT images may be segmented to produce voxel models of anatomy. Organ doses to WILLIAM from CAP and chest CT examinations can be substantially decreased by implementing TCM. References 1. Caon M (2014) The reduction in Monte Carlo calculated organ doses from CT with tube current modulation using WILLIAM, a voxel model of 7 year-old anatomy, Austral Phys Eng Sci Med, 37(4):743–752

123

O110 MCDoseCalculator – a new tool to assess doses to patient eyes during CT examination K. Offer1, A. Perdomo2, D. Hoxley3 1 Honours student, La Trobe University, Melbourne, Victoria ([email protected]). 2Radiology Department, Alfred Health, Melbourne, Victoria ([email protected]). 3Lecturer, La Trobe University, Melbourne, Victoria ([email protected])

Introduction According to recent research, the dose the eye lens can receive before radiation induced cataracts will develop is much lower than previously thought, or non-existent (Kleiman 2012). Due to advances in scanning technology, current dose estimation tools are inadequate for accurate eye lens dose estimations, and thus a new tool is needed. Method To improve on the current applications which use Monte Carlo lookup tables to estimate does, An open source GUI front end to GATE (Geant4 Application for Tomographic Emission) was written in C ++. GATE is a proven Monte Carlo simulation engine (Jan, S., et al. 2004). This new front end, named MCDoseCalculator, takes user input of CT scan parameters, runs simulations in GATE and reads the outputted doses for analysis. Calculations were performed using this tool and the ICRP reference phantoms (Menzel, Clement and DeLuca 2009), and were compared to measurements taken on a GE Lightspeed VCT with the Radcal 9095 ionization chamber (Radcal Corporation, California, USA) and values generated by CTDosimetry. Results Based on the Monte Carlo simulations within GATE, the eye lens itself received between 1 and 5 % of the overall dose to the eyeball depending on the scan parameters. This means no easy multiplication factor can be used to estimate eye lens dose based on eyeball dose. Most doses calculated were close to those measured on the Lightspeed VCT. The ability to tilt the gantry allowed for a reduction in the estimated dose to the lens of the eye of over 50 %. Conclusion The current methods used to evaluate doses to patients’ eyes have shortfalls that cannot be rectified without the use of new methods of dose calculation. MCDoseCalculator is an open source application that allows for accurate dose calculation to the eye lens, and lays a foundation for future development in the dose calculation area. References 1. Kleiman, N. J. ‘‘Radiation cataract.’’ Annals of the ICRP 41.3 (2012): 80–97. 2. Jan, S., et al. ‘‘GATE: a simulation toolkit for PET and SPECT.’’ Physics in Medicine and Biology 49.19 (2004): 4543. 3. Menzel, H. G., C. Clement, and P. DeLuca. ‘‘ICRP Publication 110. Realistic reference phantoms: an ICRP/ICRU joint effort. A report of adult reference computational phantoms.’’ Annals of the ICRP 39.2 (2009): 1.

O111 Estimating water equivalent patient thickness from DICOM data Brian Lunt1 1

MyXrayDose Ltd, Auckland, New Zealand ([email protected])

Australas Phys Eng Sci Med Introduction A quantity that represents the amount of tissue in an x-ray beam would be a more relevant measure of patient size when comparing radiographic technique and x-ray dose, than patient age, height and weight as is commonly used. Patient height and weight are not routinely recorded during diagnostic X-ray examinations, and are an indirect indication of the amount of tissue in the X-ray beam only. Within the DICOM header of a Digital Radiography image are a number of quantities including kVp, Filtration, Focus Detector Distance, Patient Entrance Dose, and image receptor Exposure Index. These quantities could be used to estimate the equivalent amount of a standard material such as water, in the X-ray beam. For this to be possible an accurate relationship between the quantities needs to be determined. Method Derivation of a set of formula to estimate water equivalent patient thickness has been undertaken by fitting equations to data generated from GEANT4 Monte Carlo simulation of a water phantom in an ideal X-ray system. Comparison of the simulated results with direct measurement using a water phantom have been undertaken to validate the simulation. Conclusion Estimation of water equivalent patient thickness from DICOM image header data using a simple set of equations is possible. Uncertainties in quantities such as Exposure Index need to be reported with any estimates. Patient size estimates in terms of water equivalent thickness are recommended as this is conceptually easy to understand and validate.

O112 Automated analysis of CT data for the purposes of optimisation

Data- or metric-based improvement efforts are important to achieving and sustaining meaningful change as a result of a quality improvement effort. As clinician-scientists, we routinely acquire and use data for research and quality assurance. Related to quality assurance, data acquired for performance evaluation is distinctly different than data acquired for quality improvement. Whereas data for performance evaluation are outcome oriented and typically consist of lagging indicators of quality and safety, data for improvement should be specific and actionable to guide quality and safety improvement efforts. The analysis of data is necessary whether data is collected for performance evaluation or for improvement. From other industries, it has be shown that control charts are an effective tool to indicate next actions when working to minimize process variation. An operational definition of quality improvement is then ensuring that the measured metric is as close to its target value as possible with minimum variation about the target value. This is the essence of process control as a method to improve quality. Process control as a quality improvement strategy is largely unutilized in the clinical radiotherapy routine. However, some vendors are starting to provide quality assurance tools that include control chart analysis as part of their software packages, which may encourage the use of process control techniques in the future. This presentation will provide a background on quality improvement, process control and control charts as well as describe how to create one type of control chart, the individuals chart. A parallel is drawn between clinical trials that would be used for any medical intervention and a quality or safety intervention. Lastly, the relationship between quality and safety and the effect of minimizing process variation on the success or failure of radiotherapy will be presented.

Andrew Blair1, Brian Lunt1 1 MyXrayDose Limited, New Zealand ([email protected]), ([email protected])

Introduction A modern CT scanner can consist of thousands of lines of protocol settings. While Diagnostic Reference Levels (DRLs) aim to optimise a small subset of these, other protocols may be rarely scrutinised. Without automated analysis it is almost impossible to check every protocol in use. In this work we present the preliminary findings of automated protocol analysis from a major tertiary hospital. Method MyXrayDose software was installed on a DICOM node within the hospital network. Each CT scanner was then configured to send all images and Radiation Dose Structured Reports to this node. DICOM header data is stripped from each image and sent to a webbased database for further analysis. Each image is also analysed to estimate patient size, water equivalent diameter, image noise, image sharpness and material composition. Results Patient size was accurately estimated for the vast majority of patients. The water equivalent diameter proved to be a more robust measure of patient size and a more useful quantity for size specific dose estimates. Image composition analyses has allowed automated anatomical localisation in some cases but further work is required. Image noise varied greatly with imaging task as expected. Conclusion Automated methods have been shown to provide an easy way of dealing with very large data sets. Data can be summarised and displayed in ways that draw attention to un-optimised protocols.

KS09 Statistical process control in radiation medicine T. Pawlicki1 1

Dept of Radiation Medicine & Applied Sciences, Univ of California San Diego, USA ([email protected])

O113 Statistical process control (SPC) for quality assurance of patient positioning during head-and-neck radiotherapy Robert J. W. Louwe1, Sarah J. Moore2 1

Wellington Blood and Cancer Centre, New Zealand ([email protected]), ([email protected]) Introduction Correct and consistent patient positioning is extremely important when highly conformal dose distributions are delivered using VMAT. However, considerable day-to-day variation in the relative positioning of anatomical structures for head and neck radiotherapy has been reported despite the use of individual head supports and fixation masks1. This study investigates the application of SPC for the quality control of patient positioning, both for individual patients and for patient groups. Method On average 7 CBCT scans per patient were registered for more than 100 Head and Neck cancer patients using each of the following match structures: C1–C3, C3–C5, C5–C7, C7-caudal, the larynx, mandible, and occipital bone. Subsequently, the patient deformation during treatment was quantified by calculating the position of each structure relative to C1–C3. Individual Value (IV) and Exponentially Weighted Moving Average (EWMA) charts were used to retrospectively analyse improvements in patient deformation for the whole patient cohort, and to evaluate the ability of SPC to monitor individual patient deformation during treatment. Results EWMA charts showed that a the magnitude and the variation of the systematic overall 3D-deformation vector decreased after the instigation of a multi-disciplinary working group from 3.0 ± 0.9 mm (1 S.D.) in 2011 to 2.2 ± 0.4 mm (1 S.D.) in the first 3 months of 2014. This improvement could be related to the implementation of a

123

Australas Phys Eng Sci Med 4

5

y = 0.0258x + 0.0852 R² = 0.7279

y = 0.0256x + 0.572 R² = 0.0837

4

heart displacement [cm]

heart displacement [cm]

different type of head rest, and re-training of staff members in maskmaking. Continued monitoring of the results from April 2014 onwards showed an apparent decline in positioning accuracy. However, this could be related to 5 exceptional cases. Application of EWMA charts to monitor individual patient deformation enabled detection of systematic changes in positioning as small as 1 mm early on in the treatment.

3 2 1

3

2

1

a.

b. 0

0 20

30 40 50 60 70 relative increase in breath hold amplitude [%]

80

0

20

40

60

80

100

120

140

relative change in combined lung volume [%]

Fig. 1 a Reported amplitude of breath hold and b relative change in combined lung volume as predictors of heart displacement in DIBH

Fig. 1 IV chart (left) and EWMA chart (right) of the systematic overall 3D - deformation vector. Red solid lines represent data from the reference period used to calculate process limits (red dashed lines). Grey and black lines represent data from the observation periods Conclusion SPC is an effective tool for reviewing and monitoring the accuracy of patient positioning. References 1. S. van Kranen et al., Setup uncertainties of anatomical sub-regions in head-and-neck cancer patients after offline CBCT guidance. Int J Radiat Oncol Biol Phys. 2009; 73(5):1566–73.

O114 Can patient breathing tracks be used to predict heart displacement in DIBH? 1

1

1

1

P. Lonski , M. Portillo , B. Chua , T. Kron 1

Peter MacCallum Cancer Centre, Melbourne, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]) Introduction Deep inspiration breath hold (DIBH) is a becoming increasingly popular as a cardiac sparing technique for patients undergoing radiotherapy for left sided breast cancer. This work investigates the use of both the patient breathing trace at CT as well as the change in lung volume as a predictor of heart displacement achieved in DIBH. Method 20 consecutive patients underwent both a conventional and DIBH CT scan on a Philips Brilliance Wide Bore scanner. CT scans for 10 of 20 patients included the whole lung and the change in lung volume was assessed for these patients as a predictor of heart displacement. The Varian RPM software was used to assess the breathing trace for the other 10 patients. The amplitude gap between quiet breathing and deep breath hold were recorded and assessed as predictors for heart displacement. To measure heart displacement, a line was drawn on the axial plane between the medial and left lateral ball bearings to define a region of interest (reference line), which typically would correspond to a high dose region. The amount of heart present from this line was measured on both scans for all 20 patients. Results The amplitude as seen on the RPM software of the breath hold did not correlate with heart displacement (Fig. 1a). The change in lung volume however was found to be a predictor for larger heart

123

displacement (Fig. 1b) from the reference line, corresponding to the high dose region. Conclusion Change in lung volume was found to correlate with heart displacement in DIBH whereas the amplitude of the breath hold did not. This may have implications for future use of spirometry for patient training to improve heart displacement. Further work is ongoing to test the feasibility of spirometry as a training tool for these patients.

O115 The impact of audiovisual biofeedback breathing guidance on thoracic 4D-CT: a digital phantom study Sean Pollock1, John Kipritidis1, Danny Lee1, Kinga Bernatowicz2,3, Paul Keall1 1 Radiation Physics Laboratory, Sydney Medical School, The University of Sydney ([email protected]). 2Center for Proton Therapy, Paul Scherrer Institute. 3Department of Physics, ETH Zu¨rich

Purpose Irregular breathing motion has a deleterious effect on thoracic and abdominal four-dimensional computed tomography (4DCT) image quality. The breathing training system: audiovisual biofeedback (AVB) has demonstrated to significantly improve the regularity of breathing motion. The purpose of this study was to quantify the impact of AVB on thoracic 4D-CT image quality utilising the digital eXtended Cardiac Torso (XCAT) phantom driven by patient anatomic motion patterns. Methods Internal and external breathing motion from 5 lung cancer patients with tumour motion [5 mm was synchronised with the XCAT phantom to simulate cine-mode and respiratory-gated 4D-CT acquisitions. Patient motion data was obtained from high-frequency MRI scans under two breathing conditions (1) without guidance (free breathing (FB)), and (2) with guidance (AVB). The following patient data was collected to synchronise with the XCAT phantom: SI and AP tumour motion, Si diaphragm motion, AP chest motion, and phase information. 4D-CT image quality was quantified by artefact presence measured using the normalised cross coefficient of adjacent slices across couch positions, in addition to the comparison between simulated and ground truth images in terms of the mean square error (MSE) intensity difference. Results Compared to FB, AVB reduced 4D-CT artefacts by 11 % (p = 0.02) in cine mode, and by 1 % (p = 0.77) in respiratory-gated mode. Compared to FB, AVB reduced MSE by 30 % (p \ 0.001) in cine mode, and by 35 % (p \ 0.001) in respiratory-gated mode. Conclusion This was the first study to quantify the impact of breathing guidance on 4D-CT image quality and image artefacts. Results demonstrated significant improvements from the use of AVB

Australas Phys Eng Sci Med for both cine mode and respiratory-gated mode acquisitions. These findings could have considerable implications for AVB as a simple method to improve the quality of 4D-CT imaging and patient treatment planning.

O117 Comparing imaging modality (CT or MRI) and treatment position (prone or supine) for breast radiotherapy inter-observer variation

O116 Characterisation and prevention of data loss due to undersampling in retrospective low-pitch helical CT (4DCT)

E. M. Pogson1,2,3, G. Delaney2,4, V. Ahern5, M. Boxer2, C. Chan6, S. David8, M. Dimigen6, J. A. Harvey8, E.-S. Koh2,3,9,10, K. Lim2, G. Papadatos2, M. L. Yap2,3,9,10,11, V. Batumalai2,3,9, E. Lazarus6, J. Shafiq3,9,10, G. Liney2,3,12, C. Moran6, P. Metcalfe12,13, L. Holloway4,12,13,14

C. Stanton1, R. Artschan1, C. Dempsey1,2, J. Lehmann1,3 1

Department of Radiation Oncology, Calvary Mater Newcastle, Australia ([email protected]), ([email protected]). 2School of Health Sciences, University of Newcastle, Newcastle, Australia ([email protected]). 3Institute of Medical Physics, University of Sydney, Sydney, Australia ([email protected]) Introduction Retrospectively correlated 4DCT scans attained using low-pitch helical methods are susceptible to data loss from undersampling. Tending to occur following reduction in patient respiratoryrate during scanning, this missing data can be hard to detect and can affect tumour delineation accuracy. To characterise data loss due to undersampling and prevent its clinical impact within the radiotherapy setting, the method of Smith et al. (2012) was applied [1]. Method A radio-opaque rod was affixed under the couch-top of a Toshiba Aquilion 16-slice CT-scanner, angled obliquely to scan direction. A CIRS dynamic thorax phantom, programmed with sinusoidal breathing motion of adjustable time period, was positioned on the couch above the rod. 4DCT scans were performed utilising Varian’s RPMTM system which samples the patient’s breath-rate for optimisation of pitch prior to scanning and co-registers the breathing signal during image reconstruction. A reduction in breath-rate from sample acquisition was introduced by increasing the time period, after acquisition and before CT scanning, mimicking a patient slowing their breathing. Visualisation of rod breakages in the coronal image plane, corresponding to missing data due to undersampling, was used to evaluate the drop in breath-rate as a function of initial acquisition value before data loss occurred. Manually reducing helical pitch from the sample acquisition value was investigated as a resolution. Results Using the radio-opaque rod allowed easy detection of missing data with enough sensitivity to detect a single missing slice in any phase bin. A drop in respiratory rate of 1–2 bpm following initial acquisition caused data loss due to undersampling. Manually reducing the acquisition value by 5 bpm resolved this by providing pitch contingency. Conclusion Characterisation of data loss due to undersampling was used to establish pitch contingency to prevent its occurrence during 4DCT-simulation. However, reducing pitch increases total scan time, temporal image blur and patient dose. References [1] Smith, D., C. Dean, and J. Lilley (2012). ‘‘A practical method of identifying data loss in 4DCT.’’ Radiotherapy and Oncology, 102(3):393–398

1 Centre for Medical Radiation Physics, Faculty of Engineering and Information Sciences, University of Wollongong, Australia. 2 Liverpool and Macarthur Cancer Therapy Centres, NSW, Australia ([email protected]), ([email protected]), ([email protected]). 3Ingham Institute for Applied Medical Research, Sydney, NSW, Australia ([email protected]), ([email protected]). 4SWSCS, University of New South Wales, Australia ([email protected]). 5Crown Princess Mary Cancer Care Centre, Westmead Hospital, NSW, Australia ([email protected]). 6Department of Radiology, Liverpool Hospital, NSW, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]). 7 Peter MacCallum Cancer Institute, Melbourne, VIC, Australia ([email protected]). 8Princess Alexandra Hospital, QLD, Australia ([email protected]). 9University of New South Wales, Sydney, NSW, Australia ([email protected]). 10Collaboration for Cancer Outcomes Research and Evaluation (CCORE), Liverpool Hospital, Liverpool, NSW, Australia ([email protected]), ([email protected]). 11University of Western Sydney, Sydney, NSW, Australia ([email protected]). 12 Centre for Medical Radiation Physics, University of Wollongong, Australia. 13Liverpool & Macarthur Cancer Therapy Centres & Ingham Institute, Liverpool, Australia ([email protected]). 14 Institute of Medical Physics, University of Sydney, Australia ([email protected])

Introduction This study compares contouring Magnetic Resonance Imaging (MRI) to Computed Tomography (CT) for breast radiotherapy treatment volumes. Methods A non-contrast T2-weighted MRI and radiotherapy CT scan were undertaken for 34 breast cancer patients in prone and supine positions. 11 observers (2 Radiologists and 9 breast Radiation Oncologists) contoured the MRI and CT data-sets. A gold standard STAPLE volume was generated from each dataset from all contours. This was compared with individual observer contours to generate Dice similarity coefficients (DSC). Additional seroma contour analysis was performed on a subset of this data (8 observers for 18 patients) as absent and axillary volumes were excluded. Results Average whole breast kappa statistics for CT and MRI were 0.81 ± 0.01 for both supine data-sets and 0.84 ± 0.01 for both prone data-sets. No measureable difference between contour conformity was seen between CT and MRI whole breast volumes, however prone DSC and kappa statistics were significantly higher than supine for both CT and MRI whole breast volumes (p \ 0.001). Prone volumes were larger than their supine counterparts (p \ 0.001). Initial seroma

123

Australas Phys Eng Sci Med analysis over all 34 patients and 11 observers had similar DSC for all data-sets. The seroma volumes (8 observers with 18 patients) showed supine CT seromas to have higher DSC than MRI supine (p \ 0.05). There was no difference between CT supine and prone seromas. Both prone and supine CT datasets had larger seromas than their corresponding MRI (p \ 0.001). The MRI prone seroma datasets were more concordant than MR supine datasets (p \ 0.05).

Table 1 The DSC and volumes (cc) for all 34 patient data-sets and 11 observers for whole breast volumes and seroma volumes. Seroma volumes for 18 patients (8 observers) are also displayed CT Supine Whole breast DSC (11 observers, 34 patients)

MRI Supine

CT Prone

MRI Prone

0.894 ± 0.005 0.892 ± 0.005 0.914 ± 0.004 0.916 ± 0.004

Seroma DSC (11 0.533 ± 0.035 0.502 ± 0.037 0.558 ± 0.033 0.521 ± 0.038 observers, 34 patients) Seroma DSC (8 observers, 18 patients)

0.719 ± 0.034 0.653 ± 0.051 0.716 ± 0.034 0.700 ± 0.045

Whole breast Vol. (11 observers, 34 patients)

684 ± 32

669 ± 33

797 ± 37

793 ± 39

Seroma Vol. (11 observers, 34 patients)

18 ± 3

14 ± 3

20 ± 3

15 ± 3

Seroma Vol. (8 observers, 18 patients)

27 ± 5

22 ± 5

28 ± 6

22 ± 5

concept of effective atomic number. The accuracy of the technique was investigated for liquid samples of known density and composition; aqueous ethanol and salt solutions. Method CT scans were conducted with near mono-energetic radiation of 30–100 keV at the Australian Synchrotron Imaging and Medical Beam Line (IMBL). The instrumentation used a medical flat panel array with CsI:Tl converter and 0.2 mm pitch, 200 9 20 mm beam and ADC ion chamber to monitor the air kerma rate. Results The radiation dose delivered by each scan was CTDIvol 5–17 mGy and DLP 4–40 mGy.cm, a fraction of that for medical CT. Reconstruction used filtered back projection with a ramp filter to 0.2 mm pixel size. For individual slices the noise to signal ratio (NSR) was 3.2–8.4 %. Analysis used the mean of 40 summed slices with an NSR of 0.8–1.9 %. The measurements were combined to obtain coefficients describing the compositional dependence of elemental cross-sections. DEXA considered all 34 materials and 45 permutations of beam energies separated by 5 to 70 keV. The difference between DEXA results and true values improved with wider energy separations reaching approximately 0.5 % (one standard deviation) for separations 20 keV or more. Propagation of errors analysis was employed to quantify contributions from random and systematic errors, accounting for the observed accuracy of the technique. Conclusion The applications for DEXA are sample characterisation and predicting interaction coefficients at other photon energies for attenuation correction and radiation dosimetry calculations.

Conclusion Prone whole breast volumes are more conformal and larger in volume than supine. Prone and Supine MRI whole breast volumes are comparable to CT and may be utilized in an MRI-only setting such as an MRI-Linac. Seroma volumes are similar, with MRI supine volumes having lower concordance (DSC) than CT supine and MRI prone volumes. Acknowledgements This work is supported by a grant from Cancer Australia and the National Breast Cancer Foundation. Fig. 1 DEXA results for 35–55 keV

O118 Dual energy X-ray analysis using synchrotron computed tomography at 30–100 keV S. M. Midgley1, N. Schleich2, A. W. Stevenson3 1

School of Physics, Monash University, Clayton, VIC 3080 Australia ([email protected]). 2Department of Radiation Therapy, University of Otago, Wellington, New Zealand ([email protected]). 3Imaging & Medical Beam line, Australian Synchrotron, CSIRO Materials Science & Engineering, Private Bag 33, Clayton South, VIC 3169, Australia ([email protected]) Introduction Dual energy X-ray analysis (DEXA) uses computed tomography (CT) measurements at two photon energies to characterise the density and composition of materials (Midgley 2004, 2011). Results are expressed as the electron density (Ne) and fourth statistical moment (R4) describing the elemental composition similar to the

123

Keywords Tissue characterisation, Kv imaging, Dosimetric comparison, Physics, Computer Applications-General, Chemistry, Acknowledgments This research was undertaken on the Imaging and Medical Beamline at the Australian Synchrotron, Victoria, Australia. The Royal Society of New Zealand is gratefully acknowledged for providing a travel grant to conduct the experiment. References 1. Midgley SM (2004) A parameterisation scheme for the x-ray linear attenuation coefficient and energy absorption coefficient. Phys. Med. Biol. 49:307–25 2. Midgley SM (2011) A model for multi-energy x-ray analysis. Phys. Med. Biol. 56:2943–62

Australas Phys Eng Sci Med Experimental, CT-Quantitative, Absorptiometry/Bone densiometry (Libre office finds 291/300 words) Author Biographical details (42/50 words) The author has found employment with the non-destructive testing industry and clinical departments providing Nuclear Medicine, PET, Radiology and Radiotherapy services. This presentation concerns research conducted outside of normal working duties, presently with the Department of Medical Imaging at Flinders Medical Centre.

O119 Development of an image guidance protocol for micro beam radiation therapy for the imaging and medical beamline at the Australian Synchrotron D. Pelliccia1,2, C. M. Poole1, J. Livingstone2, D. Ha¨usermann2, J. C. Crosbie1 1

School of Applied Sciences, RMIT University, Melbourne, Victoria, Australia. 2Imaging and Medical Beamline, Australian Synchrotron, Clayton, Victoria, Australia ([email protected]), ([email protected]). Imaging and Medical Beamline, Australian Synchrotron, Clayton, Victoria, Australia ([email protected]), ([email protected]), ([email protected]) Introduction Microbeam radiation therapy (MRT), using X-rays from a synchrotron, is a novel, preclinical form of radiotherapy that shows promise of providing a major advance in cancer control if successfully translated to clinical practice. To generate MRT, the filtered synchrotron beam is segmented by a collimator into a lattice of microbeams, usually 25–50 lm wide, spaced at regular intervals of 200–400 lm (Brauer-Krisch et al., 2010; Crosbie et al., 2010). Typical radiation doses are 300–1000 Gy (peak dose), and 5–20 Gy (valley dose), delivered in milliseconds. Method We describe the development and implementation of a protocol for image guidance for the MRT station of the Imaging and Medical Beamline (IMBL) at the Australian Synchrotron. The purpose of image guidance in clinical radiotherapy is to guarantee precise control of the radiation field to accurately deliver the prescribed dose to the target and not its surroundings. As such, a valid protocol must be able to generate live images of the patient and register these with pre-existing treatment plan. Data obtained from image guidance will inform decisions about patient positioning. Results The double-crystal Laue monochromator at IMBL produces a 20 mm vertical offset between the diffracted monochromatic beam (for imaging) and the transmitted pink beam (for treatment). Either beam can be selected by moving a slit, without changing the beam filtration or monochromator settings. In-vacuo filtering is chosen to select the treatment beam with a mean energy of 95 keV. The monochromator selects an imaging energy of 50 keV. Real time images can be registered to exiting CT-scans or directly used to align the sample, with a semi-automatic user interface. After imaging, the sample and the relevant beamline components are translated into the pink beam for treatment. Conclusion Preliminary results indicate that the proposed approach is viable and permits fast imaging, alignment and irradiation of targets with very good accuracy.

Fig. 1 Left radiography of a plastinated mouse head, acquired at the MRT station. Right CT slice, acquired in real-time of the same sample, at the position marked by the red dashed line in the left image

References 1. Brauer-Krisch, E. et al. (2010) Mutat. Res. 704, 160 (2010). 2. Crosbie, J.C. et al. (2010) Int. J. Radiat. Oncol. Biol. Phys. 77, 886.

O120 The EclipseTM treatment planning system for microbeam radiotherapy trials at the Australian synchrotron C. M. Poole1, P. A. W. Rogers2, J. C. Crosbie1 1

School of Applied Sciences, RMIT University, Melbourne, Victoria, Australia ([email protected]), ([email protected]). 2University of Melbourne Department of Obstetrics & Gynaecology, and Royal Women’s Hospital, Parkville, Victoria, Australia ([email protected]) Introduction Before clinical trials of synchrotron microbeam radiotherapy (MRT) on humans can occur, a computerised treatment planning system (TPS) to calculate the dose distribution in the patient must be developed and validated. To satisfy this requirement, we use a research licenced version of the EclipseTM TPS from Varian Medical Systems. This research license allows for customised dose calculation algorithms to be integrated with the clinical work-flows in EclipseTM that are typical to modern radiotherapy. In this way, all the functionality of the EclipseTM TPS familiar to many radiation oncology professionals is retained for application to MRT. Method Our TPS is designed for the dynamic MRT modality that has been developed for the Imaging and Medical Beamline (IMBL) at the Australian Synchrotron. It uses a pencil beam convolution algorithm for dose calculation, and allows for the design of customised conformal masks. For the treatment itself, the white beam is collimated to a 30 mm wide and 1 mm high field which illuminates the MRT collimator, which in turn produces 50 lm wide vertical microbeams separated at 400 lm center-to-center. The sample and a mask is then dynamically swept through this array of microbeams, producing a dose of radiation in the sample that is conformal to the mask aperture. Results Dose calculation considers the sample geometry derived from conventional CT data, custom bolus structures, standard and custom conformal masks, and multiple fields. Digitally reconstructed radiographs are also produced for online image guidance with portal and

123

Australas Phys Eng Sci Med CT imaging. Beam configurations and sample stage motions are limited so as to reflect the actual limits on the beamline. Conclusion We have compared the output from the MRT TPS to measurements on the beamline and documented our experiences in using it for planning the delivery of known doses to samples.

KS10 On-line adaptations for MRI guided Radiotherapy Bas Raaymakers1 1

Dep. of Radiotherapy, UMC Utrecht, The Netherlands ([email protected])

O121 Absolute dosimetry on a dynamically scanned beam for synchrotron microbeam radiotherapy J. E. Lye1, P. D. Harty1, D. J. Butler2, J. C. Crosbie3, A. W. Stevenson4 1

Australian Radiation Protection and Nuclear Safety Agency (ARPANSA), Yallambie, Victoria, Australia ([email protected]). 2Australian Radiation Protection and Nuclear Safety Agency (ARPANSA), Yallambie, Victoria, Australia. 3RMIT University, Melbourne, Australia. 4Imaging and Medical Beamline, Australian Synchrotron, Victoria, Australia Introduction The absolute dose delivered with a dynamically scanned beam in the Imaging and Medical Beamline (IMBL) on the Australian Synchrotron was measured with a primary standard graphite calorimeter. The calorimetry was compared to measurements using a free-air chamber (FAC), a TN31014 Pinpoint ionization chamber, and a 34001 Roos ionization chamber. Method The IMBL beam height is limited to approximately 1 mm. To produce clinically useful beams of a few centimetres the beam must be scanned in the vertical direction. In practice the patient/ detector is scanned and the scanning velocity defines the delivered dose. The calorimeter, FAC, and Roos chamber measure the dose area product (DAP) which is then converted to peak dose with the scanned beam area derived from film measurements. The Pinpoint chamber measures the central axis dose directly and does not require beam area measurements. The calorimeter and FAC measure dose from first principles. The ionization chambers are secondary standards calibrated in primary standard laboratories with kilovoltage X-ray tubes. The calorimeter also measures the DAP of the microbeams. Film is used to provide a high resolution relative measure of the microbeam profiles and derive the peak to valley dose ratio (PVDR). Results The calorimetry shows agreement of the order of 2 and 5 % with the FAC and ion chambers respectively. PVDR of greater than 100 were measured with a 2-shot film technique utilizing a high resolution scanning microscope (Fig. 1). Conclusion The graphite calorimeter provides an absolute measurement of the dose delivered with a dynamically scanned beam on the IMBL at the Synchrotron, and has been validated against FAC and ion chamber measurements. High resolution film measurements complete the chain to measure the microbeam peak and valley doses.

The advent of hybrid MRI radiotherapy systems provides direct anatomical feed back from the treatment table. That is, immediately prior, but also during radiation delivery, the latest state of the anatomy can be visualized. This feed-back brings also the responsibility to adapt the treatment if the anatomy has changed relative to the treatment planning. And MRI can reveal anatomical changes, due to tumor regression, changes in bowel filling, (irregular) breathing patterns or even cardiac motion. For the MRI linac, we are working towards real-time plan adaptations. The first step towards this goal is on-line plan adaptation. A rather simple on-line plan adaptation is what we named ‘‘Virtual Couch Shift’’. Here not the patient is moved to account for rigid motion, but instead the IMRT dose distribution is moved. Other options for once daily plan adaptation is of course full on-line re-planning. We are investigating automatic IMRT planning for this purpose. In order to account also for intra-fraction anatomical changes, we have developed a adaptive sequencer that fully re-optimizes the remaining treatment after each segment, while accounting for the accumulated dose. Finally within the consortium that has been formed for clinical introduction of the MRI linac, also work is on-going for real-time full plan re-optimization.

IS10 Back to the Future: some synergies between physics and medicine from history to horizon D. Thwaites1 1

Director of the Institute of Medical Physics, The University of Sydney, Australia Abstract not yet supplied

O122 An Australian mining boom: developing an Australian network for datamining routine radiotherapy clinical and radiomics data for clinical decision support David Thwaites1, Lois Holloway2, Michael Bailey3, Samir Barakat4, Martin Carolan3, Geoff Delaney2, Matthew Field4, Gary Goozee2, Joerg Lehmann5, Tim Lustberg6, Andrew Miller3, Johan van Soest6, Shalini Vinod2, Sean Walsh6, Andre Dekker6 University of Sydney ([email protected]). 2Liverpool Hospital. 3Wollongong Hospital. 4University of Sydney. 5University of Sydney (and Newcastle Mater Hospital). 6MAASTRO Clinic, The Netherlands Introduction Large amounts of routine radiotherapy (RT) data are available, which can potentially add clinical evidence to support better decisions. A developing collaborative Australian network, with a leading European partner, aims to validate, implement and extend European predictive models (PMs) for Australian practice and assess

1

Fig. 1 Microbeams measured with two different microscopes and a flatbed scanner (dashed line)

123

Australas Phys Eng Sci Med their impact on patient decisions. Wider objectives include: developing multi-institutional rapid learning, using distributed learning approaches; and assessing/incorporating radiomics information into PMs. Methods Two initial stand-alone pilots were conducted; one on NSCLC, the other on larynx, using patient datasets in two different centres. Open-source rapid learning systems were installed, for data extraction and mining to collect relevant clinical parameters from routine databases. The European DSSs were learned (‘‘training cohort’’) and validated against local data sets (‘‘clinical cohort’’). Further NSCLC studies are underway in three more centres to pilot a wider distributed learning network. Initial radiomics work is underway. Results For the NSCLC pilot, 159/419 patient datasets met the PM criteria, and hence eligibility for inclusion in the curative clinical cohort (for the larynx pilot, 109/125). Some missing data were imputed using Bayesian methods. For both, the European PMs successfully predicted prognosis groups, but with some differences in practice reflected. For example, the PM-predicted good prognosis NSCLC group was differentiated from a combined medium/poor prognosis group (2YOS 69 vs. 27 %, p \ 0.001). Stage was less discriminatory in identifying prognostic groups. In the predicted good prognosis group two-year overall survival was 65 % in curatively and 18 % in palliatively treated patients. Conclusion The technical infrastructure and basic European PMs support prognosis prediction for these Australian patient groups, showing promise for supporting future personalized treatment decisions and potentially improved practice changes. The early indications from the distributed learning and radiomics pilots strengthen this. Improved routine patient data quality should strengthen such rapid learning systems. Additional collaborators V.Ahern/R.Alvandi/M.Ebert/K.Foo/D.Fraser/A.George/A.Ghose/S.Greenham/F.Hegi/N.Kadaan/T.Kron/J.Ludbrook/C.Oberije/D.Stirling/J.Sykes/S.Yau. References 1. A Dekker, S Vinod, L Holloway, C Oberije, A George, G Goozee, G Delaney, P Lambin, D Thwaites. ‘‘Rapid Learning in Practice: A Lung Cancer Survival Decision Support System in Routine Patient Care Data.’’ Radiotherapy and Oncology 113 (2014), 47–53. doi: 10.1016/j.radonc.2014.08.013.

O123 A distributed data mining network infrastructure for Australian radiotherapy decision support Matthew Field1, Mohamed Samir Barakat1, Michael Bailey2, Martin Carolan2, Andre Dekker3, Geoff Delaney4, Gary Goozee4, Lois Holloway4, Joerg Lehmann1, Tim Lustberg3, Johan van Soest3, Jonathan Sykes5, Sean Walsh3, David Thwaites1 University of Sydney ([email protected]). 2Illawarra Cancer Care Centre. 3MAASTRO Clinic. 4Liverpool Hospital. 5 Westmead Cancer Therapy Centre 1

Introduction Automated electronic storage of medical records and imaging is becoming standard practice in radiotherapy. The aim of this study is to facilitate the improvement of radiotherapy decision support systems that may be derived from these databases. A distributed learning infrastructure that links prediction models across NSW radiotherapy clinics by sharing parameters and statistics to protect data privacy and thus base the decisions on broader sources of evidence has been established. Method The computer assisted theragnostics (CAT) system, as developed by our international collaborators at MAASTRO clinic,

was used for compiling a standardized and de-identified clinical and imaging databases [1]. An Australian research cloud computing server was used to host a distribution program written as a Java web service. This service was designed to send a compiled algorithm to each clinic and allow the algorithm to then send and receive parameters between each clinic. The algorithm implemented was a support vector machine (SVM) using the alternating direction method of multipliers for combining parameters in a consensus prediction model. Results The CAT system for lung cancer diagnosis was installed in three clinics, Liverpool, Illawarra and Westmead Cancer Care Centre with an unfiltered cohort of 4577, 1882, 1422 patients respectively. The distribution infrastructure was tested with simulated data points. The result was confirmed by repeating the algorithm when centralizing this simulated data. The accuracy of prediction on a held-out test data set was higher when using the consensus prediction model versus each localized model. Conclusion With the collaboration of medical professionals and computer scientists it is feasible to facilitate prediction models based on large cohorts of patients across multiple clinics while satisfying the data privacy rights of the patients. Using this system we intend to construct decision support systems with a range of distributed algorithms. References 1. Dekker, Andre, Shalini Vinod, Lois Holloway, Cary Oberije, Armia George, Gary Goozee, Geoff P. Delaney, Philippe Lambin, and David Thwaites. ‘‘Rapid Learning in Practice: A Lung Cancer Survival Decision Support System in Routine Patient Care Data.’’ Radiotherapy and Oncology 113, no. 1 (October 2014): 47–53. doi:10.1016/j.radonc.2014.08.013.

O124 Dynamic treatment delivery characteristics of four Elekta linear accelerators S. Arumugam1, D. P. Truant2, J. Begg1, A. Xing1, A. George1, G. Goozee1 1 Department of Medical Physics, Liverpool and Macarthur Cancer Therapy Centre and Ingham Institute, Liverpool and Campbelltown Hospital, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]). 2Department of Medical Physics, Liverpool and Macarthur Cancer Therapy Centre, Liverpool and Campbelltown Hospital, Australia ([email protected])

Introduction Volumetric Modulated Arc Therapy (VMAT) is a complex radiotherapy technique. The purpose of this study was to compare the dynamic dose delivery performance of four Elekta Linear accelerators (linacs) across two radiotherapy centres. Method Two Elekta Versa-HD and two Elekta Synergy linacs with Agility MLC heads were beam matched across two centers. The dynamic delivery of 6MV beam was characterised and compared by the following tests1, 2: 1. Leaf position accuracy in rotational delivery. 2. Flatness and symmetry constancy at 10 different dose rates and constant gantry speed. 3. Dose constancy with variable leaf speeds (2.5 mm/s to 35 mm/s). 4. Dose with variable leaf gap (5 to 25 mm). 5. Dose constancy with:

123

Australas Phys Eng Sci Med – – – –

Different dose rates and constant gantry speed. mutual variation of gantry speed and doserate. mutual variation of leaf speed and doserate at constant gantry speed. mutual variation of leaf speed, dose rate and gantry speed.

Results The figure below shows an example of the planned, EPIDmeasured and Dynalog-recorded values for (a) MU, (b) gantry speed, (c) dose rate and (d) MLC position as a function of gantry angle.

6. Dosimetric impact of leaf reversal in dynamic delivery. The dose measurements were performed using a gantry mounted ion chamber 2D-array. Expected dose for the tests 3–5 were calculated using a 6 MV beam model in Pinnacle3 planning system. Results Mean(r) leaf positional accuracy across all machines is within 0.2(0.7)mm. Overall difference in Flatness, Symmetry and Dose constancy of the beam in arc mode at different dose rates is within 1 %. Dose constancy with variable leaf gap and speed is within 2 % of the Pinnacle3 calculation. Dose constancy with mutual variation of dose rate, leaf sped and gantry speed is within 2 % of the Pinnacle3 calculated values for the range of speeds and dose rates studied. Maximum under-dose caused by the reversal of leaves at maximum speed is in the order of 3 %. Conclusion The dynamic delivery performance of all four linac is within the published recommendations1,2. Minimal inter-machine variations were observed. References 1. ’’Code of Practice for the Quality Assurance and Control for Volumetric Modulated Arc Therapy’’ (2015) Netherlands Commission on Radiation Dosimetry. 2. D. G. Kaurin, L. E. Sweeney, E. I. Marshall and S. Mahendra, ‘‘VMAT testing for an Elekta accelerator’’. Journal of Applied Clinical Medical Physics 13 (2) (2012).

Conclusion We have demonstrated that an EPID frame grabber system can be used for efficient time-resolved linac QA for VMAT. The measurements show good agreement with Dynalog files and DICOM plan files. Future work will be to perform these tests periodically to systematically verify using EPID and Dynalog files for VMAT QA. References 1

STRALINGSDOSIMETRIE, N. C. V. (2015). NCS Report 24: Code of practice for the quality assurance and control for volumetric arc therapy. 2 Van Esch, A. et al. (2011). Implementing RapidArc into clinical routine: a comprehensive program from machine QA to TPS validation and patient QA. Medical Physics 38(9), 5146–5166. 3 Ling, C. et al. (2008). Commissioning and quality assurance of RapidArc radiotherapy. International Journal of Radiation Oncology* Biology* Physics* 72(2), 575–581

O126 Lung SABR dose calculation verification using a convolution superposition algorithm O125 Time-resolved linac QA for VMAT: A comprehensive and efficient system using an EPID frame grabber B. Zwan1,2, J. Hindmarsh1, E. Seymour1, K. Sloan1, R. David1,2, M. Barnes3, C. Lee1,2 Central Coast Cancer Centre, CCLHD, Gosford, NSW. 2School of Mathematical and Physical Sciences, University of Newcastle, Newcastle, NSW, Australia ([email protected]). 3 Calvary Mater Hospital, Newcastle, NSW, Australia 1

Introduction Volumetric Modulated Arc Therapy (VMAT) relies on the synchronisation of three dynamic components; gantry angle, monitor units (MU) and MLC-defined beam shape. The accuracy of these components should be assessed individually and as a function of time (or gantry angle) to ensure correct dose delivery.1Many VMAT QA schedules do not adequately measure the performance of all dynamic linac components on a routine basis.2, 3In this work, a set of tests have been developed to monitor Varian linac performance during VMAT deliveries using an EPID frame grabber system. Method Test VMAT plans have been designed to assess the following machine parameters as a function of gantry angle: (1) MLC position, (2) MLC speed, (3) MLC acceleration, (4) MLC-gantry synchronisation, (5) dose rate, (6) gantry speed, (7) flatness/symmetry, (8) mechanical sag and (9) MU-gantry synchronisation. The above parameters were measured by acquiring MV image frames every 100 ms using an EPID frame grabber. Image processing techniques were used to extract the MLC positions, MU, mechanical sag and flatness & symmetry from each frame. The parameters were also extracted from machine log files (Dynalog files) and the DICOM plan file for comparison.

123

N. Hardcastle1, B. Oborn2, A. Haworth3 1

Northern Sydney Cancer Centre, Royal North Shore Hospital, NSW, Australia ([email protected]). 2Illawarra Cancer Care Centre, NSW, Australia ([email protected]). 3 Peter MacCallum Cancer Centre, VIC, Australia ([email protected]) Introduction Stereotactic Ablative Body Radiotherapy (SABR) is an established treatment technique for primary lung tumours and metastases. SABR aims to deliver an ablative, highly conformal dose to the tumour in five or fewer fractions. Dose calculation and subsequent verification in lung SABR is challenging due to large tissue density variations and small fields. A new software (Mobius3D) designed to perform dose calculation verification using an advanced dose calculation algorithm is evaluated for lung SABR. Method Ten lung SABR plans spanning a range of treatment volumes and locations were selected. The treated 3D conformal treatment plans were performed in the XiO treatment planning system (TPS) using the superposition algorithm. Verification of the planned dose was performed with Mobius3D. For comparison, the plans were recalculated in Eclipse TPS using the AAA algorithm, as well as an in-house BEAMnrc Monte Carlo algorithm. Dosimetric parameters relating to target and organ at risk dose were compared between the calculation algorithms. A Wilcoxon signed-rank test was used to compare each method with a threshold of significance of p \ 0.05. Results Calculated point doses in the tumour were greater with Mobius3D, Monte Carlo and RadCalc compared with XiO by up to 12.5, 5.5 and 18.0 % respectively. Near minimum target doses were greater with AAA, Mobius3D and MC compared with XiO by up to 8 %, and near maximum doses were greater than XiO by 11.0 % and

Australas Phys Eng Sci Med 6.5 % respectively. AAA, Mobius3D and MC all calculated higher lung V5 Gy and V20 Gy. All stated results were statistically significant. Conclusion In the ten lung SABR cases considered, significant differences between the TPS and alternate calculation methods were identified. Identifying the true dose in these situations is challenging and clinical judgement is required to evaluate the comparison of results of the TPS and plan verification software.

O127 STAT RAD – an expedited scan-plan-QA-treat workflow for single fraction bone metastases SBRT L. L. Handsfield1, D. Wilson2, Q. Chen3, P. Read2 1

Medical Physics, Auckland City Hospital, NZ ([email protected]). 2Radiation Oncology, University of Virginia, Virginia, USA (DDW3 [email protected]), ([email protected]). 3Medical Physics, University of Virginia, Virginia, USA (QC3 [email protected]) Introduction We set out to develop a highly conformal yet also more clinically efficient and patient-centric palliative radiotherapy workflow. Radiation Oncology guidelines1 advise against the use of extended fractionation schemes for the palliation of bone metastases and state that strong consideration should be given to single fraction treatments. Database review of treatments from 2005 to 2011 found single fraction treatment still only make up a small percent of the typically prescribed bone metastases radiation prescriptions (3–8 % in the USA and 17 % in Australia)2. Method A Monte Carlo secondary dose calculation and a phantomless, exit-detector based plan-specific QA method were developed to expedite the pre-treatment QA process. A Failure Mode and Effects Analysis (FMEA) was performed on the proposed new workflow to ensure safe clinical introduction. To allow for efficient workflow, one physicist was dedicated to creating and quality assurance testing the plan, while another physicist preformed a secondary check of each aspect before treatment. Results The FMEA resulted in several changes to the proposed workflow, including a comprehensive checklist to be completed throughout the procedure, retaining the kV treatment planning CT, and altering the originally proposed quality assurance plans. The resulting workflow averaged 3 h from the time the patient enters the CT room to the end of treatment (CT simulation: 30 min; MD contour and review: 60 min; treatment planning: 30 min; plan documentation and QA: 30 min; treatment delivery: 30 min). Conclusion We have successfully implemented a 3 h, highly conformal, single fraction palliative radiation therapy workflow for patients with painful advanced bone metastases. Our process does require a high level of coordination between the radiation therapists, physician, planner, and physicist, but does not significantly interrupt the daily clinical flow. Our workflow may be preferred by the patients and may result in a savings of time and money for the department. References 1. C. Hahn, B. Kavanagh, A. Bhatnagar, G. Jacobson, S. Lutz, C. Patton, L. Potters, M. Steinberg, ‘‘Choosing wisely: the American Society for Radiation Oncology’s top 5 list,’’ Pract. Radiat. Oncol. 4, 349–355 (2014). 2. C.E. Rutter, J.B. Yu, L.D. Wilson, H.S. Park, ‘‘Assessment of national practice for palliative radiation therapy for bone metastases suggests marked underutilization of single-fraction regimens in the United States,’’ Int. J. Radiat. Oncol. Biol. Phys. 91, 548–555 (2015).

O128 Achievable dose gradients in spinal radiotherapy treatments delivered via Tomotherapy T. Kairn1,2, S. Harris3, Z. Moutrie4, S. B. Crowe2,3 1 Genesis Cancer Care Queensland, Brisbane, Australia. 2Queensland University of Technology, Brisbane, Australia ([email protected]). 3Royal Brisbane and Women’s Hospital, Brisbane, Australia ([email protected]), ([email protected]). 4Genesis Cancer Care NSW, Crows Nest, Australia ([email protected])

Introduction The spinal cord is the major dose-limiting organ for spinal radiotherapy, due to its location, surrounded by vertebral structures and potentially abutting or penetrating the desired treatment volume. Safe treatment of a vertebral target with a tumoricidal radiation dose therefore requires a steep dose gradient between the targeted vertebra and the spinal cord. This study investigates whether such dose gradients can be reliably produced using a helical Tomotherapy system. Method Six sample treatments were planned for delivery to vertebral and paraspinal targets in the thoracic and lumbar spine. A prescription of 30 Gy in 10 fractions was used for all targets, as this is reportedly the most common non-SBRT prescription used in spinal radiotherapy [1]. To minimise the risk of myelopathy [2], the planned cord dose was constrained to less than 21 Gy. Gafchromic EBT2 film measurements in a water equivalent phantom [3] were used to verify that the dose gradients predicted by the treatment planning system were clinically deliverable. Results The figure presents a sample dose distribution (left), sagittal dose plane from film measurement (centre) and comparison of planned and measured dose profiles (right, with profile location shown in centre). In this example, where the treatment volume encompasses the whole vertebra and surrounds the spinal cord, the dose gradient in the spinal canal was 17 cGy/mm (5.3 %/mm) according to the plan and 22 cGy/mm (6.7 %/mm) according to the measurement. In simpler cases, where the target did not entirely surround the cord, gradients between 4.8 and 5.2 %/mm were observed, with the gradient in the measurement slightly exceeding the gradient in the plan, in all cases.

Conclusion Planned dose distributions met planning constraints in all six cases. Film measurements showed the predicted doses to be accurate, including precipitous dose gradients at the edges of the spinal canals. References 1. Sahgal, A., et al. (2011) J. Neurosurg. Spine 14: 151–166 2. Schultheiss, T. E., et al. (1995) Int. J. Radiat. Oncol. Biol. Phys. 31(5): 1093–1112 3. Kairn, T. et al. (2011) Australas. Phys. Eng. Sci. Med. 34(3): 333–343

123

Australas Phys Eng Sci Med

O129 Impact of Interplay Effect on the Dosimetry of SBRT Treatments using 6MV FFF beams J. Moorrees1, C. Stanton1, J. Simpson1, J. Lehmann2 1 Department of Radiation Oncology, Calvary Mater Newcastle, Newcastle, Australia ([email protected]), ([email protected]), ([email protected]). 2Institute of Medical Physics, University of Sydney, Sydney, Australia ([email protected])

Introduction VMAT has seen increased demand for SBRT treatments in recent years, especially in tumours which experience intrafraction motion. In these cases the interplay effect may have a significant effect on the dose distribution within the CTV, the extent of which depends on a number of factors including organ motion (amplitude, phase, and period), number of arcs, dose-rate and MLC modulation. The impact of these factors on the dose distribution of a moving target using Varian RapidArc was investigated in order to provide guidance on patient selection and treatment planning within the Department. Method A low (\3 MU/cGy), medium (3–4 MU/cGy) and high modulation ([4 MU/cGy) lung plan was re-planned for a single fraction SBRT treatment using the Varian Eclipse treatment planning system with a 6-MV FFF beam (1400 MU/min) and either two, three or four arcs in the optimisation. The two arc plans were then delivered to a CIRS dynamic thorax phantom with a moving lung insert of cos4 motion, amplitude of 0.5, 1 and 2 cm, and breathing periods of 4, 6 and 8 s. Each of the above deliveries was completed for four different starting phases. Following the method of Ong et al. (2011, 2013), these dynamic deliveries were compared to a static delivery which was convolved with the motion function and a gamma pass rate calculated [1, 2]. Results It was found that plans with a smaller range of motion, higher period and increased number of arcs reduced the interplay effect, the latter due to the averaging effect from starting at a random breathing phase. Conclusion For plans with expected high modulation it is suggested that patients should be appropriately selected which have breathing amplitudes and periods below certain threshold values. Increasing the number of arcs reduces the clinical impact of the interplay effect. References [1] Ong, C., W. Verbakel, and J. Cuijpers et al. (2011). ‘‘Dosimetric impact of interplay effect on RapidArc lung stereotactic treatment delivery.’’ Int. J. Radiation Oncology Biol. Phys., 79(1):305–311 [2] Ong, C., M. Dahele, B. Slotman, and W. Verbakel (2013). ‘‘Dosimetric impact of the interplay effect during stereotactic lung radiation therapy delivery using flattening filter-free beams and volumetric modulated arc therapy on RapidArc lung stereotactic treatment delivery.’’ Int. J. Radiation Oncology Biol. Phys., 86(4):743–748

O130 Dose discrepancy in spine SBRT: A case report S. Arumugam1, M. Berry1, C. Ochoa1, A. Xing1, P. Vial2 1 Department of Medical Physics, Liverpool and Macarthur Cancer Therapy Centre and Ingham Institute, Liverpool and Campbelltown Hospital, Australia ([email protected]), ([email protected]),

123

([email protected]), ([email protected]). 2Department of Medical Physics, Liverpool and Macarthur Cancer Therapy Centre and Ingham Institute, Liverpool and Campbelltown Hospital, Australia ([email protected]) Introduction Stereotactic Body Radiation Therapy (SBRT) to treat spinal metastases has shown excellent clinical outcomes for local control. High dose gradients wrapping around spinal cord make this treatment technically challenging. In this work we present a spine SBRT case where a dosimetric error was identified during pre-treatment dosimetric Quality Assurance (QA). Method A patient with metastasis in T7 vertebral body consented to undergo SBRT. The Clinical Target Volume (CTV) consisted of the entire vertebral body, pedicle, ipsilateral transverse process, and ipsilateral lamina. A dual arc VMAT plan was generated on the Pinnacle treatment planning system (TPS) with a 6 MV Elekta machine using gantry control point spacing of 4˚. Standard pre-treatment QA measurements were performed, including ArcCHECK, ionchamber in CTV and Spinal cord (SC) region and film measurements in multiple planes. An observed dose discrepancy in the SC region was investigated by repeat measurements, replanning with different TPS settings, and comparing plan complexity indices. Results The table shows the plan complexity index (LT-MCSv) and dose measurement results of original and regenerated plans:

Plan

Point dose ArcCHECK LTMCSv agreement (%) gamma pass rate CTV

SC

2 %G/ 2 %L/ 2 mm 2 mm

Original plan

0.2044 0.9 %

21.1

88.5

84.5

Repeat plan

0.2043 1.2

16.6

89.0

85.1

-2.3

91.1

87.8

Repeat plan-3˚ CP spacing 0.2079 -0.4

All plans had similar complexity scores. The dose measured to the SC was significantly higher than reported by TPS in the original and repeat plans. Acceptable agreement was only achieved when the gantry control point spacing was reduced to 3˚. The ArcCHECK measurements showed clinically acceptable pass rate in all plans. Conclusion A potentially harmful dose error was identified by pretreatment QA. TPS parameter settings used safely in conventional treatments should be re-assessed for complex treatments. Dose verification checks using multiple detector systems is warranted to ensure the safety of complex treatments.

O131 Implementation of a single database for the management of all equipment QC data at Wellington Blood and Cancer Centre R. A. Day1, L. Greig1 1 Wellington Blood and Cancer Centre, Wellington, New Zealand ([email protected]), ([email protected])

Introduction The Wellington Blood and Cancer Centre (WBCC) operates a quality assurance program involving up to 370 separate equipment quality control (QC) tests per month. In 2013 a project was

Australas Phys Eng Sci Med initiated to improve the management of QC data and facilitate better compliance with regulatory codes of safe practice (NZ Ministry of Health, 2003, 2010). These codes require evidence of an equipment QC program that has a well-defined review and reporting structure. Compliance with the codes is externally audited. A decision was made to consolidate all equipment QC into a single database, to eliminate hardcopy and uncontrolled spread sheets of QC data and to replace an aging QC database. Method The database software selected needed to accommodate QC data from treatment, imaging, planning, and measurement equipment. Key requirements were flexibility to build customized QC tests, functionality for formal QC data review, functionality for internal and external audit, ability to interface with third party measurement equipment, and use of an industry standard database format. From various commercial and in-house options, the SNC AtlasTM software was selected. The project to transfer equipment QC to the new database took close to 2 years. New processes were introduced where QC data has to be formally reviewed by a senior staff member within set timescales. Monthly internal and annual external QC audits are conducted. Results All WBCC equipment QC data is now searchable in one database with full traceability of who conducted and reviewed each test, and when. The commercial database solution chosen met many of the needs, but the reporting functionality was found to be lacking, and several software bugs were discovered in the first software versions. Conclusion The move to a single database for all radiotherapy equipment QC has improved processes and traceability. The commercial solution chosen did not fully meet all the needs.

However when using patient image sets these errors are rarely known to a reasonable degree of accuracy. Mathematical phantoms such as XCAT allow introduction of known setup errors and well changes to the anatomy, with the drawback of typically being geometric models describing structure boundaries and not incorporating observed features in patient images such as tissue inhomogeneity, motion effects, and statistical uncertainty inherent to the imaging modality. Work has been done addressing realistic simulation of CT imaging via Monte Carlo modelling (Segars et al. 2008, Tabary et al. 2009), however these still use uniform tissues. This work describes a time and resource efficient method for simulating these effects based on image processing. Method Attenuation coefficient phantoms of the thorax were modelled using XCAT2 incorporating breathing and cardiac motion cycles and were then converted to arrays of Hounsfield units. Five equally time spaced frames spanning 0.2 s starting at end-expiration were generated and averaged to simulate time averaging effects. A Gaussian filter was applied to smooth the resulting stepped edges of structures. Tissue inhomogeneities were simulated by applying salt and pepper noise and then expanding the singular salt and pepper voxels using a Gaussian filter. Finally, random noise with a Gaussian distribution was applied to simulate the statistical uncertainty in imaging. Various combinations of anatomic and noise simulation parameters were investigated and evaluated by radiologists. Results Figure 1 shows examples of coronal slices of simulated CT images of the same thoracic XCAT phantom using different noise simulation parameters.

References 1. New Zealand Ministry for Health (2010), Code of safe practice for the use of irradiating apparatus in medical therapy (CSP12), version 1.4. 2. New Zealand Ministry for Health (2003), Code of safe practice for the use of sealed radioactive materials for brachytherapy (CSP14), version 1.4. 3. New Zealand Ministry for Health (2010), Code of safe practice for the use of x-rays in medical diagnosis (CSP5), version 1.3. 4. New Zealand Ministry for Health (2010), Code of safe practice for the use of unsealed radioactive materials in medical diagnosis, therapy and research (CSP3), version 1.3. Acknowledgements QC test developers and testers: V. Aguas, J. Donaldson, R. Louwe, I. Ramadaan, T. Satherley, B. Scarlet, N. Schleich, J. Steel, B. Steer, A. Williams. Wellington Blood and Cancer Centre, Wellington, New Zealand

Fig. 1 Example coronal slices of simulated CT images of the same XCAT phantom using different noise simulation parameters Conclusion Simulated x-ray CT images were produced from XCAT attenuation coefficient outputs via image processing, without requiring time and resource expensive Monte Carlo modelling. Such images could be used for a range of applications including IGRT QA and assessment of image registrations. References

O132 Simulated x-ray CT images for IGRT quality assurance utilising XCAT phantoms J. R. Supple1, S. Siva2, T. Kron1,3, M. L. Taylor1, R. D. Franich1 1 School of Applied Sciences and Health Innovations Research Institute, RMIT University, Melbourne, VIC, Australia ([email protected]), ([email protected]), ([email protected]). 2Department of Radiation Oncology, Peter MacCallum Cancer Centre, East Melbourne, VIC, Australia ([email protected]). 3Physical Sciences, Peter MacCallum Cancer Centre, East Melbourne, VIC, Australia ([email protected])

Introduction Error trapping in CT-based IGRT can be achieved with image sets incorporating known setup errors or anatomy changes.

1. Segars W, Mahesh M, Beck T, Frey E and Tsui B 2008 Realistic CT simulation using the 4D XCAT phantom Med. Phys. 35:3800–3808 2. Tabary J, Marache-Francisco S, Valette S, Segars W P and Lartizien C. Realistic X-Ray CT simulation of the XCAT phantom with SINDBAD. Nuclear Science Symposium Conference Record (NSS/MIC), 2009 IEEE, 2009. IEEE, 3980–3983.

O133 A novel device with sub millimetre resolution for the measurement of mechanical and radiation isocenter in external beam radiotherapy Paul Archer1, Prabhakar Ramachandran1 1

PeterMac ([email protected]), (Prabhakar [email protected])

123

Australas Phys Eng Sci Med Introduction Over time various methods have been used to measure the mechanical isocenter as a function of the three primary axis of rotation on isocentric linear accelerators, additionally measurement of the radiation isocenter using the Winston–Lutz methodology is often the preferred method. The objective is to demonstrate through the use of one instrument the measurement of isocenter for the three mechanical axes of rotation and related radiation isocenters, with the capability to reference these measurement sets with respect to each other. Method A device has been designed and built to measure and record with sub millimetre precision the three mechanical rotational axes (Collimator, Couch and Gantry). With minimal hardware and setup changes the same instrument employs the Winston- Lutz technique to generate data points for the radiation isocenter. Data points from the mechanical measurements are collected from an electronic sensor and processed before being sent to a computer application written in C++ for analysis and graphical representation. Dicom image data from the Winston Lutz tests can be imported from the electronic portal imager into the software application. Results The data points from the mechanical axis measurements and the Winston Lutz images are then displayed by the software application either independently or simultaneously, with the capability to overlay and match the recorded mechanical and radiation isocenters this facilitates and assists in identifying discrepancies. The software provides graphical representation of the isocenter shape and size. Conclusion The device prototype is currently undergoing internal assessment and has returned results with sub millimetre precision; the system is designed for use on Varian linear accelerators; however the device can be adapted for use on linear accelerators from other vendors or telecobalt machines.

O134 Evaluation of the TrueBeam Machine Performance Check (MPC) Beam Constancy Checks M. Barnes1,2, R. Artschan1, J. Lehmann1,2, P. Greer1,3 1

Calvary Mater Newcastle, Newcastle, Australia ([email protected]). 2School of Medical Radiation Sciences, The University of Newcastle, Newcastle, Australia ([email protected]). 2Institute of Medical Physics, The University of Sydney, Sydney, Australia ([email protected]). 3School of Physics, The University of Newcastle, Newcastle, Australia ([email protected]) Introduction Machine Performance Check (MPC) is a Varian provided integrated suite of image based linear accelerator (linac) QA tests designed to be used for daily QA on TrueBeam linacs. MPC checks include beam constancy checks acquired from single EPID acquisitions per beam energy without the IsoCal phantom in the beam. The beam constancy checks include beam output change, beam uniformity change and beam centre shift relative to user defined baseline images. Method MPC was run daily concurrently with departmental standard daily QA tests on the TrueBeam linac for a period of 3 months. Additionally, the TrueBeam monthly QA program was accelerated to fortnightly. The MPC beam constancy checks were compared to Daily QA3; output, flatness and symmetry and field shift checks as well as to fortnightly ion chamber outputs, IC profiler profiles and in-house EPID based; output, profile constancy and EPID panel positional checks for all four clinical beams (6 MV, 6 MV FFF, 10 MV and 10 MV FFF). Results The MPC output constancy check showed good agreement to both daily and monthly standard in-house procedures. MPC output

123

constancy was found to be sensitive to an output adjustment of 1.2 %. Due to the large number of linac variables inherent in both the MPC uniformity (beam steering, beam energy and constancy of EPID pixel response) and beam centre shift (focal spot position, jaw calibration and EPID panel position) checks it proved difficult to correlate MPC results to standard QA checks as recommended by best practice protocols such as AAPM TG-142. Conclusion MPC is a fast efficient beam checking tool. The output constancy check is accurate, but the beam uniformity constancy and beam centre shift results are difficult to interpret in terms of standard QA test parameters. References 1. Clivio, A., Vanetti, E., Rose, S., Nicolini, G., Belosi, M., Cozzi, L., Baltes, C., & Fogliata, A. (2015). ‘‘Evaluation of the Machine Performance Check application for TrueBeam linac.’’ Radiation Oncology 10(97): XXX. 2. Klein, E., Hanley, J., Bayouth, J, Fang–Fang, Y., Simon, W., Dresser, S., Serago, C., Aguirre, F., Ma, L., Arjomandy, B., & Liu, C. (2009) Task Group 142 report: Quality assurance of medical accelerators. Med. Phys. 36: 4197–4212.

O135 A custom Quasar phantom for real-time measurements of linear accelerator time delay in gated treatments A. M. C. Santos1, J. Shepherd1 1

School of Physical Sciences, University of Adelaide, Adelaide, Australia, and Department of Medical Physics, Royal Adelaide Hospital, Adelaide, Australia ([email protected]) Introduction Respiratory gated treatments are becoming more common in order to reduce motion uncertainties. One issue associated with gated treatments is the time delay between the gating system and the linear accelerator. In this study we develop and characterise an affordable phantom to be used in routine QA of the Varian Real-Time Position Management (RPM) system. Diodes have been incorporated into the phantom in order to estimate the time delay. Method A commercial Quasar phantom was customised to incorporate two stepper motors which independently control an anteriorposterior abdomen moving plate, and an inferior-superior moving lung insert. Various lung inserts were made including an ion chamber and film insert. A photodiode placed in the path of the radiation is used to measure when beam on and beam off occurs. Two Arduino microcontroller boards have been utilised to control the motors, read the diodes and write to an SD card. The program developed to control the Arduino boards allows for a patients breathing trace to be input to the phantom. Results Figure 1 shows an example of a triangular wave function which the phantom produced. By superimposing the RPM measured breathing trace of the phantom with the motion that was intended, a good agreement is observed indicating that the phantom is operating as expected. By then comparing the diode signal with the gated output signal from the RPM system, little time delay is observed between the RPM system and the linac. This time delay is of the order of 0.1–0.2 s. Conclusion An affordable moving phantom has been developed for routine QA of respiratory gating systems. It is capable of mimicking patients breathing traces and performing real-time dose-rate measurements with diodes. Results indicate a time delay of the order of 0.1–0.2 s for the RPM system.

Australas Phys Eng Sci Med Method A 5 mm ball phantom is positioned at the isocentre of the CBCT system through iterative imaging and position adjustment. Following this, the position of the ball relative to the treatment isocentre is assessed using MV cine images acquired with an arc exposure using a 30 mm stereotactic cone and with the gantry rotating throughout its range. Couch rotation verification is done similarly with the gantry pointing down and the couch rotating throughout its range. The MV cine images are analysed using previously published methods (Rowshanfarzad, Sabet et al. 2011). Results Over a nine-month period the max deviation of isocentre coincidence measured with the MV panel ranged from 0.25 to 1.29 mm for gantry rotation and 0.52 to 1.54 mm for couch rotation. The consistent monitoring prompted corrective action through adjustment of isocentre agreement and communication with radiation oncologists to ensure with adequate CTV-PTV margin selection. Assessment of the data over this period shows some user variability. Data from the new TrueBeam linac shows tighter results compared to the Trilogy, with max deviations ranging up to 0.74 mm. Conclusion A Winston-Lutz style test based on automated analysis of MV cine-images is feasible and in our practice helped to identify the need for corrective action. A software improvement could be to display deviation from ideal MV position allowing for non-perfect positioning of the ball phantom. References 1. Rowshanfarzad, P., M. Sabet, D. J. O’Connor and P. B. Greer (2011). ‘‘Verification of the linac isocenter for stereotactic radiosurgery using cine-EPID imaging and arc delivery.’’ Med Phys 38(7): 3963–3970.

Fig. 1 The custom Quasar phantom results with [above] the RPM measured and the phantom designated positions, and [below] the RPM gating times and the real-time diode measurements

O136 Medium term experience with automated linac based Winston-Lutz style testing of gantry and couch rotations R. Jones1, R. Artschan1, M. Doebrich1, C. Stanton1, B. Aldrich1, M. Barnes1, P. Greer1, J. Lehmann1,2 1 Calvary Mater Newcastle, Newcastle, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]). 2The University of Sydney, Sydney, Australia ([email protected])

Introduction Accurate image-guided stereotactic radiotherapy requires precisely coincident imaging and treatment isocentres. Winston-Lutz style tests are commonly used to verify such agreement. Our Winston-Lutz test technique involves iteratively aligning a spherical ball-bearing test object to the CBCT isocentre and then imaging the ball-bearing with the MV beam, establishing the relative isocentre positions. We report results of nine-months of regular Winston-Lutz measurements on a Varian Trilogy linac and on the recent, short-term results of a Varian TrueBeam.

P01 Deconstructing treatment planning system and beam model accuracy: Wedges, asymmetry, inhomogeneites and the ACDS Level II audit A. D. C. Alves1, S. Keehan1, J. Lye1, L. Dunn1, J. Lehmann1, J. Kenny1, M. Shaw1, I. Williams1 1 Australian Clinical Dosimetry Service, Yallambie, VIC, ([email protected])

Introduction Each level of the ACDS’ tiered system of audits evaluates different parts of the dosimetric chain in radiation therapy. The Level I and Ib audits evaluate the absorbed dose under the facility’s reference dosimetry conditions. The Level III audit is an end-to-end test of the treatment chain performed on-site using an anthropomorphic phantom. The Level II audit lies between these two extremes focusing solely on the treatment planning system dose calculation. Method The Level II audit measures planar dose using an ionization chamber array positioned at multiple depths in rectilinear homogeneous and inhomogeneous phantoms composed of solid water and lung. Computer generated data sets of the phantoms are supplied to the facility for planning a range of cases including reference fields, asymmetric fields, and wedged fields. The audit assesses 3D planning with 6 MV photons and a 0 gantry. Scoring is performed using local dose differences between the planned and measured dose within 80 % of the field width. The overall result is determined by the maximum dose difference over all scoring points, cases, and planes. Pass (Optimal Level) is B3.3 %, Pass (Action Level) is B5.0 %, and Fail (Out of Tolerance) is [5.0 %. Results The ACDS has performed 38 Level II audits. 21 of the audits passed (optimal), 6 passed (action level), 9 failed, and the remaining audits was not assessable. The high fail rate is largely due to a

123

Australas Phys Eng Sci Med systemic issue with modelling asymmetric 60 wedges which caused a delivered overdose of 5–8 %. Conclusions the ACDS has demonstrated the powerful diagnostic ability of the Level II audit and has rigorously tested the treatment planning systems implemented in Australian radiotherapy facilities. The large data set also provides a useful benchmark. Recommendations from audits have led to facilities modifying clinical practice and changing planning protocols.

P02 An assessment of nursing staffs’ knowledge of radiation protection and practice M. K. Badawy1, K. S. Mong2, P. L. U2, P. Deb1 1

Department of Medical Sciences, RMIT University, Australia ([email protected]), ([email protected]). 2 Department of Medical Physics, Austin Health, Australia ([email protected]), ([email protected]) Although the exposure to nursing staff is generally lower than the allowable radiation worker dose limits, awareness and overcoming fears of radiation exposure is essential in order to perform routine activities in certain departments. Furthermore, the nursing staff, whether they are defined as radiation workers or not, must be able to respond to any radiological emergencies and provide care to any patient affected by radiation. This study aims to gauge the awareness of radiation safety among the nursing staff at a major hospital in different departments and recommend if further radiation safety training is required. A prospective multiple choice questionnaire was distributed to 200 nurses in 9 different departments. The questionnaire tested knowledge that would be taught at a basic radiation safety course. 147 nurses (74 %) completed the survey with the average score of 40 %. Furthermore, 85 % of nurses surveyed felt there was a need for radiation safety training in their respective departments to assist with day to day work in the department. An increase in radiation safety materials that are specific to each department is recommended to assist with daily work involving radiation. Moreover, nursing staff that interact with radiation on a regular basis should undertake radiation safety courses before beginning employment and regular refresher courses should be made available thereafter.

P03 Patient radiation dose assessment using different dose metrics and radiation dosimeters in dental cone beam Computed Tomography M. S. Bakkari1, K. M. Soliman1, A. M. Alenezi1 1

Medical Physics Department, Prince Sultan Military Medical City, Riyadh, Saudi Arabia ([email protected]), ([email protected]), ([email protected]) Introduction the objectives of the presented work was to compare different radiation dose metrics or index and dosimeters while evaluating patient doses from a dental cone beam Computed tomography (DCBCT) scanner. By using several radiation dose metrics it is possible to accurately estimate the patient effective dose (ED) simply by calculating conversion factors (Cf) relating the specific dose metric used with the ED.

123

Therefore, it is possible to compare the practicality of the most commonly used radiation dose metrics in DCBCT dosimetry today and recommend the most practical in our opinion. Method Gafchromic XR-QA2 films were used to measure the Entrance skin dose (ESD), the Dose area product (DAP) and the peak skin dose (PSD) and dose profile (DP) during Dental Cone Beam Computed Tomography (CBCT) examinations using female adult anthropomorphic phantom. We have also measured the computed tomography dose index (CTDI) using CTDI pencil ionisation chamber both free in air CTDIair and using CTDI head phantom CTDIw. MOSFET detectors were used to measure the patient eye dose by placing them directly over the eyes of the anthropomorphic female phantom and compared with the calculated doses using PCXMC 2.0 software. Results We have calculated conversion factor relating the measured entrance surface dose (ESD) using radiochromic films to organ doses calculated using PCXMC software. The measured ESD values for the most common dental CBCT examinations performed in our hospital and the PSD are reported. The obtained results seem to agree with other published dosimetric studies in the literature. Conclusion According to our study, Gafchromic films (XR-QA2) were found to be an efficient and reasonably accurate dosimetry method that can easily be implemented clinically to measure the entrance surface doses and dose profiles during irregular radiation pattern and shape like the ones encountered in dental CBCT investigations. References 1. Araki K, Patil S, Endo A, Okano T (2013) Dose indices in dental cone beam CT and correlation with dose-area product. Dentomaxillofac Radiol. 2013 May; 42(5): doi: 10.1259/dmfr.20120362. 2. Endo A, Katoh T, Vasudeva S, Kobayashi I, Okano T (2013) A preliminary study to determine the diagnostic reference level using dose-area product for limited-area cone beam CT. Dentomaxillofac Radiol. 2013 Apr;42(4): doi: 10.1259/dmfr.20120097. 3. Morant JJ, Salvado´ M, Herna´ndez-Giro´n I, Casanovas R, Ortega R, Calzado A (2013). Dosimetry of a cone beam CT device for oral and maxillofacial radiology using Monte Carlo techniques and ICRP adult reference computational phantoms. Dentomaxillofac Radiol. 2013:42(3): doi: 10.1259/dmfr/92555893. 4. Pauwels R, Theodorakou C, Walker A, Bosmans H, Jacobs R, Horner K, Bogaerts R (2012) Dose distribution for dental cone beam CT and its implication for defining a dose index. Dentomaxillofac Radiol. 2012 Oct;41(7):583–93. 5. Koivisto J, Kiljunen T, Tapiovaara M, Wolff J, Kortesniemi M (2012) Assessment of radiation exposure in dental cone-beam computerized tomography with the use of metal-oxide semiconductor field-effect transistor (MOSFET) dosimeters and Monte Carlo simulations. Oral Surg Oral Med Oral Pathol Oral Radiol. 2012 Sep;114(3):393–400. 6. Morant JJ, Salvado´ M, Casanovas R, Herna´ndez-Giro´n I, Velasco E, Calzado A (2012) Validation of a Monte Carlo simulation for dose assessment in dental cone beam CT examinations. Phys Med. 2012 Jul;28(3):200–209. 7. Han S, Lee B, Shin G, Choi J, Kim J, Park C, Park H, Lee K, Kim Y (2012). Dose area product measurement for diagnostic reference levels and analysis of patient dose in dental radiography. Radiat Prot Dosimetry. 2012 Jul;150(4):523–31. 8. Podnieks EC, Negus IS (2012). Practical patient dosimetry for partial rotation cone beam CT. Br J Radiol. 2012 Feb; 85(1010):161–167.

Australas Phys Eng Sci Med

P04 Examination of the correlation between patientdependent parameters and radiation dose rates measured around patients undergoing PET/CT imaging using 18F-FDG M. S. Bakkari1, K. M. Soliman1, S. A. Alqahtani2, H. F. Alnufaie1, A. M. Alenezi1 1

Medical Physics Department, Prince Sultan Military Medical City, Riyadh, Saudi Arabia ([email protected]), ([email protected]), ([email protected]), ([email protected]). 2Radiodiagnostic and Medical Imaging Department, Prince Sultan Military Medical City, Riyadh, Saudi Arabia ([email protected])

Introduction Patients undergoing 18F- FDG PET/CT imagings are external sources of radiation. Accurate dose rate estimates is important for conducting realistic risk assessments and performing dose reconstruction in cases of accidental exposures. The patient radiation self –attenuation factor is assumed to be a function of the patient’s body size metrics; but can we use these metrics to predict the dose rate around the patients with accuracy? The objectives of this work were to measure the patient attenuation factor by performing direct dose rate measurements from patients, to study the possible correlation between the measured dose rate constant from the patients and their body size metrics and to measure the patients’ voiding factor. Five different body size metrics were tested for correlation in this study. Method The measured dose rate at voiding time divided by the activity calculated at the voiding time and corrected for radioactive decay was used for each patient to calculate the dose rate per unit activity constant. Radiation dose measurements were done immediately before and after voiding, in order to calculate the dose rate reduction factor. The radiation dose rate was measured using a calibrated ionisation chamber. The FDG dose was administered using a calibrated automatic dose injector. Results We have measured dose rates at one meter from 57 patients and found an average dose rate of 92.2 ± 14 lSv.h-1 GBq-1. There was no statistically significant correlation between the dose rate constant per unit activity and the patient body size metrics. The measured patient voiding factor was found to be equal to 0.89 ± 0.06. Conclusion the dose rate constant of 92 lSv.h-1.GBq-1 proposed by the AAPM, TG-108 report is adequate for radiation protection purposes. The presented data can be used by medical physicist working in nuclear medicine in formulating more accurate risk estimations resulting from radiation exposure from patients undergoing 18F-FDG PET/CT imaging. References 1. Madsen MT, Anderson JA, Halama JR, Kleck J, Simpkin DJ, Votaw JR, Wendt III RE, Williams LE, Yester MV (2006). AAPM Task Group 108: PET and PET/CT Shielding Requirements. Medical Phys; 33(1):4–15. 2. Hays MT, Segall GM (1998). A mathematical model for the distribution of Fluorodeoxyglucose in humans. J Nucl Med;40:1358–1366 3. Jones SC, Alavi A, Christman D, Montanez I, Wolf AP, Reivich M (1982).The Radiation Dosimetry of 2-[F-18]Fluoro-2-Deoxy-DGlucose in Man. J NucI Med;23: 613–617. 4. Mejia AA, Nakamura T, Masatoshi I, Hatazawa J, Masaki M, Shoichi W (1991). Estimation of absorbed doses in humans due to intravenous administration of Fluorine-18-Fluorodeoxyglucose in PET studies. J Nucl Med; 32:699–706.

5. Dowd MT, Chen CT, Wendel MJ, Faulhaber PJ, Cooper MD (1991). Radiation dose to the bladder wall from 2-[18F] fluoro-2deoxy-D-glucose in adult humans. J Nucl Med; 32:707–712. 6. Cho IH, Han EO, Kim ST (2014). Very different external radiation doses in patients undergoing PET/CT or PET/MRI scans and factors affecting them. Hell J Nucl Med; 17(1): 13–18. 7. Watson CC, Casey ME, Bendriem B, Carney JP, Townsend DW, Eberl S, Meikle S, DiFilippo FP (2005). Optimizing Injected Dose in Clinical PET by Accurately Modeling the Counting-Rate Response Functions Specific to Individual Patient Scans. J Nucl Med; 46:1825–1834 8. Halpern BS, Dahlbom M, Auerbach MA, Schiepers C, Fueger BJ, Weber WA, Silverman DHS, Ratib O, Czernin J (2005). Optimizing imaging protocols for overweight and obese patients: a lutetium orthosilicate PET/CT study. J Nucl Med; 46(4):603–607. 9. Halpern BS, Dahlbom M, Quon A, Schiepers C, Waldherr C, Silverman DH, Ratib O, Czernin J (2004). Impact of patient weight and emission scan duration on PET/CT image quality and lesion detectability. J Nucl Med; 45(5):797–801. 10. Masuda Y, Kondo C, Matsuo Y, Uetani M, Kusakabe K (2009). Comparison of imaging protocols for 18F-FDG PET/CT in overweight patients: optimizing scan duration versus administered dose. J Nucl Med; 50(6):844–848. 11. Zeff, BW, Yester MV (2005). Patient self-attenuation and technologist dose in positron emission tomography. Med Phys; 32(4):861–865. 12. Quinn B, Holahan B, Aime J, Humm J, St-Germain J, Dauer LT (2012). Measured dose rate constant from oncology patients administered 18F for positron emission tomography. Medical Physics; 39(10):1071–79. 13. Yi Y, Stabin MG, McKaskle MH, Shone MD, Johnson AB (2013). Comparison of measured and calculated dose rates near nuclear medicine patients. Health Phys; 105(2):187–191.

P05 Compare dosimetric characteristics for integrated and cine acquisition modes of a-Si EPID Omemh Bawazeer1,2,3, Sisira Herath2, Siva Sarasanandarajah1,2, Tomas Kron2, Pradip Deb1 1 Discipline of Medical Radiations, School of Medical Science, RMIT University, Australia. 2Medical Physics, Peter MacCallum Cancer Centre, Australia. 3Discipline of Sciences, School of Medical Physics, Umm Al-Qura University, Saudi Arabia ([email protected])

Introduction Electronic portal imaging device (EPID) has very interesting features for dosimetry applications. The latest generation of EPID is an array of photodiode detectors on an amorphous silicon (a-Si) glass substrate (1). There are two standard image acquisition modes of using a-Si EPID that are utilized for the dosimtery purpose: integrated mode and cine mode. The aim of this study is to compare between dosimetric characteristics for integrated and cine mode. Method aS500 amorphous-silicon EPID that attached to a Varian Clinac 21iX linear accelerator (Varian Medical Systems, Palo Alto, USA) was used in this study. The EPID was irradiated with 6 &18 MV photon beam using a dose rate of 600 monitor unit (MU)/ min. Dosimetry characteristics were examined for each mode including: linearity, reproducibility, field size dependence, dose rate dependence, impact of patient body thickness, ghosting effect, changing in MLC speed and delivering intensity modulated radiotherapy (IMRT) field. Results EPID with integrated and cine modes for 6 &18 MV has comparable linear response for MU between 100 and 500 MU,

123

Australas Phys Eng Sci Med however, cine mode for both energies provide nonlinear response for small MU. When deliver IMRT plan, the subtraction between the response of integrated and cine mode give similar results for each IMRT field that is roughly equal to 1.3004 9 10 +05. Conclusion The correction for the performance of cine mode at small MU is required. Integrate and cine mode has differences in the number of frame acquired yielding differences in absolute pixel value. Even the number of frame acquired is different for each mode, the integrated and cine mode have comparable dosimteric characteristics for 6 & 18 MV. For doismteric purpose, each mode required specific calibration factor, and cine mode may require a calibration factor for each dose rate.

Results Wall material of the Blue Phantom is 10 mm thick Acrylic plastic (Perspex) with a water equivalent thickness of 11.9 mm. The uncertainty in the orthogonality of the front surface of the tank to the horizontal radiation beam axis was 0.57 (± 1 mm accuracy on the laser measurement device at 10 cm offaxis). Gantry 0 and 90 comparisons of PDD and profiles agreed within 1 and 1.5 % respectively. Conclusion The methodology described can be used to undertake dosimetry measurements with horizontal beams consistently as necessary for the Australian MRI-linac program. References

References 1. Van Elmpt W, McDermott L et al. A literature review of electronic portal imaging for radiotherapy dosimetry. Radiother Oncol. 2008;88(3):289–309.

P06 The Australian MRI-Linac program: measuring profiles and PDD in a horizontal beam J. Begg1, A. George1, S. Alnaghy2, T. Causer2, T. Alharthi3, B. Dong1, G. Goozee1, G. Liney1, L. Holloway1, P. Keall4 1 Department of Medical Physics, Liverpool and Macarthur Cancer Therapy Centre and Ingham Institute for Applied Medical Research, Sydney, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]). 2Centre for Medical Radiation Physics, University of Wollongong, Wollongong, NSW, Australia ([email protected]), ([email protected]). 3Institute of Medical Physics, School of Physics, University of Sydney, Sydney, NSW, Australia ([email protected]). 4Sydney Medical School, University of Sydney, Sydney, Australia and Ingham Institute for Applied Medical Research, Liverpool, Australia ([email protected])

Introduction The Australian MRI-Linac Program[1] is developing a 1-T open-bore MRI/6-MV linac using a Varian Linatron-MP (Palo Alto, USA) horizontal 6-MV beam. This work presents the methodology used to measure large field profiles and PDDs for a horizontal beam in a Scanditronix Wellhofer Blue Phantom (IBA-Dosimetry, Germany). Method To validate measurements made in a horizontal geometry, a CC13 ionisation chamber was used to compare PDD and profiles from an Elekta Synergy (Stockholm, Sweden) measured at gantry 0 and 90. The long axis of the chamber was orthogonal to the radiation beam axis for both geometries. For gantry 0 measurements, the watertank was set-up following standard procedures [2] and effective point of measurement, Peff [3], corrections applied in-software. For horizontal beam measurements orthogonality of the watertank front surface to the radiation beam axis was verified by comparing the distance between the reference surface of the Linac and the front surface of the tank at multiple points off-axis with a laser measurement device (Fluke, Everett, USA). Isocentre depth position was set as close to the target as the watertank side would physically allow. Corrections for the distance between the centre of the chamber at this position and the outside surface of the watertank, Peff and effective wall thickness of the watertank (twinqpl) [3] were applied post-measurement.

123

1. Keall, P. J., Barton, M., & Crozier, S. (2014, July). The Australian Magnetic Resonance Imaging–Linac Program. In Seminars in radiation oncology (Vol. 24, No. 3, pp. 203–206). WB Saunders. 2. Das IJ, Cheng CW, Watts RJ, Ahnesjo¨ A, Gibbons J, Li XA, Lowenstein J, Mitra RK, Simon WE, Zhu TC (2008) Accelerator beam data commissioning equipment and procedures: Report of the TG-106 of the Therapy Physics Committee of the AAPM. Medical physics 35:4186 3. IAEA-TRS 398: Absorbed Dose Determination in External beam Radiotherapy: An Internation Code of Practice for Dosimetry based on Standards of Absorbed Dose to Water (2000). International Atomic Energy Agency

P07 Absolute dosimetry of leksell gamma knife with solid water phantom K. Y. T. Biggerstaff1, P. H. Charles1, Chris Oliver2, Duncan Butler2 1

Radiation Oncology, Princess Alexandra Hospital, Woolloongabba, QLD Australia ([email protected]), ([email protected]). 2Australian Radiation Protection and Nuclear Safety Agency, Yallambie, NSW, Australia ([email protected]), ([email protected]) Introduction There is currently no official protocol on the calibration and quality assurance of Leksell Gamma Knife (LGK), although there is one in preparation (AAPM report 178). Several papers have reported on the use of TG-21, TG-51 and variations of each of these. Meltsner (2009) adapted TG-21 and reported to give outputs consistently 1.5–2.9 % higher than the dose calculated in the TPS system. Drzymala (2008) created a water-filled phantom and determined the dose according to TG-21, TG-51 and a combination. The three methods agree to within 1.4 %. McDonald (2011) utilizes Seuntjen’s method for in phantom calibration using the Elekta’s Solid Water LGK dosimetry phantom. Results were in agreement with Drzymala (2008). They also reported -1.8 % difference between the modified TG-51 method and TG-21 (TG-51 reports lower). This poster summarises the current methodology in detail and investigates the possibility of each correction factor being measured at a Primary Standards laboratory. Method An Exradin A1SL chamber and a IBA CC13 were used for absolute dosimetry (both chambers were calibrated in Co-60 at ARPANSA). These were placed in the solid water Elekta LGK phantom. The calibrated chamber was then used to determine absorbed dose to water by Eq. 1 where ks,w is the phantom correction factor which converts the dose to water measured in a phantom to dose to water measured in water.. Elekta determined this value via Monte Carlo simulations to be 1.005 when using solid water. Direct

Australas Phys Eng Sci Med measurements in water and the solid water phantom will be verified using the Co-60 beam at ARPANSA. Dw;Qo ¼ MQo NDw;Qo ks;w

ð1Þ

Conclusion The current status of absolute dosimetry on the LGK has been summarised. The new suggested methodology paves way for future detectors to be used for absolute dosimetry on a LGK, with individual factors for each detector and phantom directly measured by a Primary Standards laboratory, as opposed to relying on generic simulation data. References 1. American Association of Medical Physicists in Medicine (1983) Task Group 21: A protocol for the determination of absorbed dose from high-energy photon and electron beams, Med. Phys, (10) 741 2. American Association of Medical Physicists in Medicine (1999) Task Group 51: A AAPM’s TG-51 protocol for clinical reference dosimetry of high-energy photon and electron beams, Med. Phys, (26) 1847 3. Drzymala et. al. (2008) Calibration of the Gamma Knife using a new phantom following the AAPM TG51 and TG21 protocols Med. Phys, 35(2) 514 4. Johansson J, (2010) Application of a new formalism for absolute dosimetry of the Leksell Gamma Knife, Reference Number SSM 2010/2201, Project nr 4017003-06, Elekta Instrument AB 5. McDonald (2011) Calibration of the Gamma Knife Perfexion using TG-21 and the solid water Leksel1 dosimetry phantom Med. Phys, 38(3) 1685 6. Meltsner S and DeWerd L (2009) Air kerma based dosimetry calibration for the Leksell Gamma Knife Med. Phys, 36(2) 339 7. Seuntjens et al. (2005) Absorbed dose to water reference dosimetry using solid phantoms in the context of absorbed-dose protocols Med. Phys, (32) 2945

microDiamond detector were exposed to approximately 4 Gy in various phantoms. For the EPID measurements, the MUs were reduced to approximately by one tenth and images were acquired using integrated mode. The coefficient of variation (CV) was calculated for all sets of readings. Results The CV measured with EBT3, EPID and microDiamond detector were generally higher for the MLC reset experiment compared to no MLC reset. For fixed MLC aperture, the CV values varied between 0.30 and 0.34 %, 0.27–0.55 % and 0.03–0.04 % for EBT3 film, EPID and microDiamond detector. With MLC reset, the CV values were 0.33–0.69 %, 0.30–0.82 % and 0.08–0.15 %. Conclusion The results show that EBT3, EPID and microDiamond detector exhibit similar dosimetric behaviour for MLC reset experiment where CV values are generally higher compared to no MLC reset. References 1. Gotstedt J, Hauer A K and Back A Med. Phys. 42, 3911 (2015) Development and evaluation of aperture-based complexity metrics using film and EPID measurements of static MLC openings. 2. Younge KC et al. Med. Phys. 39, 7160 (2012) Penalization of aperture complexity in inversely planned volumetric modulated arc therapy. 3. Cho G and Thwaites D Med. Phys. 41 262 (2014) SU-E-T-175: Evaluation of the relative output ratio for collimator jaw and MLC defined small static 6 MV photon fields.

P09 Investigating dosimetric effects of damaged couch tops D. Binny1, C. M. Lancaster1, S. Crowe1,2

P08 An analysis of small field dosimetry uncertainty due to field aperture position variations Gwi A. Cho1,2, Joerg Lehmann2,3, David Thwaites2 1 Chris O’Brien Lifehouse at RPA, Sydney, Australia. 2Institute of Medical Physics, School of Physics, University of Sydney, Sydney, Australia ([email protected]). 3Calvary Mater Hospital, Newcastle, NSW, Australia

Introduction Small field apertures in volumetric modulated radiation treatment (VMAT) plans contribute to dosimetric uncertainty due to limitations in TPS dose calculation and MLC positioning accuracy (Gostedt et al. (2015), Younge et al.(2012)). Various detectors are used for VMAT pre-treatment QA including Gafchromic EBT3 film, electronic portal imaging device (EPID) and solid state detectors. The dosimetric response of these detectors to MLC positioning uncertainty is not well understood. This study compared the relative response of EBT3 film, EPID and microDiamond chamber for a subset of MLC and solid jaw defined small field apertures used for Eclipse TPS photon beam source modelling. Method Data was measured for a Novalis Tx linac with HD120 MLC and PortalVision aS1000 for a 6 MV photon beam. Gafchromic EBT3 film, EPID and microDiamond detector were exposed five consecutive times. Field apertures were defined by the jaws and with square MLC opening of 0.5 and 1.0 cm. This was repeated after MLC position was reset (Cho and Thwaites, (2014)). EBT3 films and the

1 Cancer Care Services, Royal Brisbane and Women’s Hospital, Brisbane, Australia ([email protected]), ([email protected]), ([email protected]). 2Faculty of Science and Engineering, Queensland University of Technology, Brisbane, Australia

Introduction The effects of treatment couch during radiotherapy is rarely fully assessed during the treatment planning process. Incorporating a couch model by fusion with the patient CT dataset into the treatment planning system (TPS) enables planners to dosimetrically evaluate the effects of beam couch intersections. However if the couch is physically damaged its CT dataset can no longer be used as an exact representation of the planned beam couch intersection due to altered dose distribution for which the magnitude estimated is often imprecise. In this study we demonstrate how existing TPS tools can be used to quantitatively predict the magnitude of a simulated damage. Method Four simulated couches were used in this study; Exact Couch IGRT(thin), BrainLab/iBeam couch with H&N extension, QFix couch and Siemens p-A couch which are existing models in the Eclipse v13.5TM TPS. A dent of an approximate volume of 2 cm3 was contoured in each of the couches and a 3D conformal planned dose to a reference point with and without the dent was noted. Planned transmission through these couched were also taken note of along with their effects on a H&N test patient plan using Volumetric Modulated Arc Therapy(VMAT). Analytical Anisotropic Algorithm (AAA) was used to calculate dose to all points.

123

Australas Phys Eng Sci Med suggest that absorbance measurements are (like conventional optical density measurements) sensitive to the post-irradiation darkening of the film and the positioning of the sample relative to the light source.

Fig. 1 Representation of a beam passing through a damaged Exact Couch IGRT Results and Conclusion Based on the analysis of the calculated point doses it was observed that the dented Exact Couch IGRT gave the highest difference when used during a 3D conformal treatment with beams passing through it compared to all the other couches in this study. VMAT deliveries through all dented couches showed no significant difference in the planned dose point due to a very small segment of the beam passing through the dent. The study substantiates that the use of a damaged couch is not recommended however its clinical implication must be assessed prior to its continued use during a daily treatment.

P010 Spectrophotometric analysis of radiochromic film S. B. Crowe1, S. Sylvander1, T. Kairn2 1

Cancer Care Services, Royal Brisbane & Women’s Hospital, Brisbane, Australia ([email protected]), ([email protected]). 2Genesis Cancer Care Queensland, Brisbane, Australia ([email protected]) Introduction The apparent colour of EBT2 and EBT3 radiochromic films has changed over time, although the manufacturer has reported no change in active-layer composition since the films were released. This study used spectrophotometry to investigate the physical basis of this anecdotal, qualitative observation. Method The CE202 spectrophotometer was used to measure absorbance of EBT2 and EBT3 film from different batches. Absorbance spectra were measured for non-irradiated samples of (a) EBT2 film from a batch supplied soon after EBT2 film’s release (expiry 2011), (b) EBT2 film from a batch supplied around the time of EBT3 film’s release (expiry 2013), (c) EBT3 film from a batch supplied soon after EBT3 film’s release (expiry 2013), and (d) a recent batch of EBT3 film (expiry 2016). Additionally, samples of recent EBT3 film were irradiated to known doses and absorbances measured at 635 nm, to provide an absorbance-dose calibration relationship. Results The observed visual changes in the colour of EBT2 and EBT3 films is supported by absorbance measurements. Comparison of the absorbance spectra for the older and newer EBT3 film samples shows that the small absorbance peaks centred around the wavelengths of orange and green light have declined in relative magnitude, for the newer EBT3 film, compared with the large peak centred near the wavelength of yellow light. Calibration results for EBT3 film

123

Conclusion Spectrophotometry measurements showed important differences between film samples from different batches and between film samples irradiated to different doses. The CE202 spectrophotometer could be used for absolute dose calibration of radiochromic film measurements, in circumstances where measurement uncertainties of 3 % are acceptable and where doses less than 800 cGy are delivered.

P011 Seeing through metal implants in gel dosimetry A. Asena1, T. Kairn1,2, S. T. Smith1, S. B. Crowe1,3, J. V. Trapp1 1 Queensland University of Technology, Brisbane, Australia ([email protected]), ([email protected]), ([email protected]). 2 Genesis Cancer Care Queensland, Wesley Medical Centre ([email protected]). 3Cancer Care Services, Royal Brisbane & Women’s Hospital ([email protected]) Introduction This study demonstrates the degradation in image quality, and subsequent dose evaluation inaccuracies, that are encountered when an optical-CT system reconstructs an image slice of a gel dosimeter containing an opaque implant, and evaluates the feasibility of a simple correction method to improve the accuracy of radiotherapy dose distribution measurements under these circumstances. Method MATLAB was used to create a number of different virtual phantoms and treatment plans along with their synthetic projections and reconstructed data sets. These simulations highlight the problems associated with using the filtered back projection code to reconstruct an image with incomplete projection data. A method of linearly interpolating the missing projection data prior to image reconstruction in order to minimise artefacts and improve dose evaluation accuracy was also evaluated. Results The results have illustrated that accurately evaluating 3D gel dose distributions in the vicinity of high-Z interfaces is not possible using the filtered back projection method, without correction, as there are serious artefacts throughout the dose volume that are induced by the missing projection data. Similar artefacts were present in physical measurements of irradiated PAGAT gel containers when read by an optical-CT system. Some of these artefacts can be minimised by increasing the number of angular projections set during image acquisition to counteract the missing projection data caused by the implant. An interpolation correction performed prior to reconstruction via the FBP algorithm has been shown to significantly improve dose evaluation accuracy to within approximately 15 mm of the opacity. Conclusion With careful placement of the implant within the gel sample, and use of the linear interpolation method described in this study, there is the potential for more accurate optical CT imaging of gels containing opaque objects.

Australas Phys Eng Sci Med

P012 Measuring small field output factors with a PTW 60018 SRS diode P. Stevenson1, D. Binny2, S. B. Crowe2, S. Sylvander2 1 Queensland University of Technology, Brisbane, Australia ([email protected]). 2Royal Brisbane & Women’s Hospital, Brisbane, Australia ([email protected]), ([email protected]), ([email protected])

Introduction Improved radiation therapy treatment options and techniques are increasingly requiring the use of small radiation fields (\3 9 3 cm2) [1] and very small radiation fields (\1.5 9 1.5 cm2) [2]. This project involved the practical application of small field dosimetry measurement procedures to establish output factors (OF) using a PTW 60018 Diode SRS for both jaw- and MLC-defined fields. Method In-tank measurements were made with the detector at 5.0 cm depth in a 6 MV beam, according to recommendations in the literature [3]. The jaw- and MLC-defined square fields sizes examined were 5 9 5, 4 9 4, 3 9 3, 2 9 2 and 1 9 1 cm2, with an additional 0.5 9 0.5 cm2 jaw-defined field included. Monte Carlo generated correction factors [4] were applied to the readings. Subsequent OF were plotted against both nominal and delivered field sizes (in terms of in-plane and cross-plane field edges, and area), for comparison with the results of other centres. Diode measurements were also evaluated against values obtained using Gafchromic EBT3 film in virtual water. Results The OF determined in the study were comparable to those measured at nearby centres and in the literature. Small differences were observed in OF between jaw- and MLC-defined fields, even where evaluated in terms of delivered field size. This difference was approximately 1 % at the smallest comparable field size of 1 9 1 cm2. Conclusion By following established procedures for small field dosimetry, and comparing results against those reported in the literature, a clinical radiation therapy department can be confident that the small field output factors obtained through in-tank measurements are accurate and suitable for input into a treatment planning system.

([email protected]). 2Medipix Team, PH-ESE-ME, CERN, Geneva, Switzerland ([email protected]), ([email protected]). 3CERN, DGS-RP, Geneva, Switzerland ([email protected]). 4Mid-Sweden University, Sundsvall, Sweden ([email protected]) Introduction The characterisation of scatter radiation in medical radiodiagnostic practices is a crucial issue for radiation safety surveillance and radiation protection optimisation as well as for the improvement of dose monitoring of medical staff. Energy-resolving hybrid pixel detectors have unique properties, which were used for the characterisation of radiation fields, and, in the future, will provide an important new technology for operational radiation protection applications. Method In this work, we used the two hybrid pixel detectors, Dosepix and Timepix, to characterise the scattered radiation field in a radiological room equipped with a CT scanner (GE HD 750 Discovery). Both systems are highly portable, with small readout systems connected to a computer through USB, and requiring no additional power supplies or cooling equipment; which make both systems practical for operational radiation protection and for radiation monitoring at medical installations. We first validated the detectors by studying their response to beam qualities (RQR3, RQR5 and RQR7) relevant for medical practices according to the International Electrotechnical Commission (IEC) and the International Standard Organisation (ISO). We then used the high-data-rate, energy-resolving capabilities of Dosepix to measure the spectra of radiation scattered in the patient examination room and at the shielded operator work station during typical clinical CT scans of a thorax phantom. In complement, we were able to use the spatial-resolving abilities of Timepix and image the CT primary beam profile, which precisely characterizes the beam width. Results For each position in the room, energy spectrum measurements were performed at three different heights; a three dimensional dose map inside the radiological room was obtained. We found that the Timepix detector can be used for the control of the beam collimation as part of the quality assurance program. Conclusion Both systems, Dosepix and Timepix, appear to be excellent tools for radiation protection survey in radiology.

References 1. Das, IJ et al. (2008). Small fields: nonequilibrium radiation dosimetry. Med Phys (35): 206–215. 2. Charles, PH et al. (2014) A practical and theoretical definition of very small field size for radiotherapy output factor measurements. Med Phys 41(4): 041707. 3. Kairn, T et al. (2015) Clinical use of diodes and micro-chambers to obtain accurate small field output factor measurements. Australas Phys Eng Sci Med 38(2): 357–367. 4. Benmakhlouf, H. (2015) Key Data for the Reference and Relative Dosimetry of Radiotherapy and Diagnostic and Interventional Radiology Beam. Doctoral Thesis in Medical Radiation Physics at Stockholm University.

P13 Measurement of scatter radiation energy spectra in CT scan rooms with the energy-resolving dosepix and timepix detectors C. Bailat1, M. Campbell2, J. Damet1,3, C. Elandoy1, E. Fro¨jd2,4, W. S. Wong2 1 Institute of radiation physics, Lausanne University Hospital, Lausanne, Switzerland ([email protected]),

P014 Comparison of Gamma pass rate for Tomotherapy DQA plan created in high and low dose region for patient specific QA using ArcCHECK Aitang Xing1, Sankar Arumugam2, Shrikant Deshpande2, Lois Holloway2,3, Gary Goozee2 1 Liverpool Cancer Therapy Centre, Liverpool, Australia ([email protected]). 2Department of Medical Physics, Liverpool Hospital, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]). 3Ingham Institute for Applied Medical Research, Liverpool, Australia

Introduction The purpose of this study was to investigate the impact of DQA plans created in high and low dose regions on the Gamma passing rates of patient specific QA results using ArcCHECK. Method Ten clinical patient plans were selected to represent the typical treatment sites in our clinic. For each patient, two DQA plans were created using the TomoTherapy DQA Station (Hi-Art version 4.2.1) on CT images of ArcCHECK phantoms. One DQA plan was created in such a way that the detector array fall in the dose region

123

Australas Phys Eng Sci Med where the dose is higher than 95 % of the prescribed dose to the patient. Another DQA plan was made to ensure the detector array in a dose region where the dose is less than 30 % of the prescribed dose. All DQA plans were delivered on the Tomotherapy Hi-Art unit in a single measurement session. The measured results were loaded into SNC (version 6.6) for comparison with the TPS-calculated dose. The Gamma index was calculated globally and locally using 3 %/3 mm, 2 %/3 mm with 10 % dose threshold of maximum TPS calculated dose. Results The averaged pass rates ± standard deviation for Gamma indices calculated using global method were (97.17 ± 1.84) % in low dose region and (97.96 ± 3.11) % in high dose region for 2 %/3 mm, (98.86 ± 1.66) % in low dose region and (99.43 ± 0.555) % in high dose region for 3 %/3 mm. On the contrast, the mean Gamma pass rate calculated locally were (87.44 ± 6.08) % in low dose region and (86.56 ± 7.07) % in high dose region for 2 %/3 mm, (91.8 ± 4.10) % in low dose region and (90.81 ± 5.73) % in high dose region for 3 %/3 mm. Conclusion The DQA plans using ArcCHECK created either in high dose region or low dose region can be used for Tomotherapy delivery quality assurance with different gamma pass rate threshold and criteria.

P015 Sensitivity evaluation of two commercial dosimeters in detecting helical Tomotherapy treatment delivery errors Shrikant Deshpande1, Mark Geurts2, Philip Vial1, Peter Metcalfe3, Lois Holloway1 1

Department of Radiation Oncology, Liverpool and Macarthur Cancer Therapy Centres and Ingham Institute, Sydney, New South Wales ([email protected]), ([email protected])., ([email protected]). 2Department of Human Oncology, School of Medicine and Public Health, University of Wisconsin- Madison USA ([email protected]). 3Centre for Medical Radiation Physics, University of Wollongong, Wollongong, Australia ([email protected]) Introduction This work assesses the sensitivity of two commercially available dosimetry systems in detecting errors in Tomotherapy delivery. Method Two commercially available dosimeters (1) MatriXX Evolution (IBA) with OmniPro-ImRT software (2) ArcCheck (Sun Nuclear) with SNC Patient software were considered. Treatment plans for ten previously treated nasopharynx Tomotherapy patients were analysed. For each patient, error plans were created by independently introducing systematic offsets in: (a) jaw width (JW) error ±1, ±1.5 and ±2 mm, (b) couch speed (CS) error ±2, ±2.5, ±3 and ±4 %, and (c) MLC leaf open time (MLC LOT) errors (3 separate MLC errors: leaf 32 open during delivery, leaf 42 open during delivery and 4 % random reductions in MLC LOT. All error plans along with the no error plan for each patient were measured using both dosimeters in the same session to minimize any machine output or experimental variations. The gamma evaluation technique (3 %/ 3 mm) was applied to quantitatively compare the measured dose from two dosimeters against the treatment planning system (TPS).The sensitivity was determined as rate of drop in gamma pass rate with increase in error magnitude.

123

Results The gamma pass rates decrease with increase in error magnitude for both dosimeters with varying sensitivity (Fig. 1). ArcCheck was insensitive for CS error up to 2.5 % and JW error of -1 mm compared to MatriXX while for MLC LOT error both dosimeters were equally sensitive.

Fig. 1 Comparison of the gamma pass rate for varying magnitude of intentional delivery error. Error bars show the standard deviation from 10 patient data measurements Conclusion MatriXX and Arccheck dosimeters showed varying sensitivity in detecting different error types and magnitudes. Both were able to pick up clinically relevant delivery errors except with ArcCheck for CS up to 2.5 % and JW up to -1 mm for this patient cohort.

P016 Feasibility of MV cine imaging for monitoring of DIBH treatments M. Doebrich1,2, J. Downie1, C. Stanton1, C. Dempsey1,2, J. Lehmann1,3 1

Calvary Mater Newcastle, Newcastle, Australia ([email protected]), ([email protected]). 2The University of Newcastle, Newcastle, Australia ([email protected]), ([email protected]). 3The University of Sydney, Sydney, Australia ([email protected]) Introduction The assessment of breath hold for left sided breast cancer treatment with deep inspiration breath hold (DIBH) can be achieved with several approaches, ranging from video based skin surface monitoring and breathing motion detection systems to the observation of a light field or a laser line on the patient’s skin. The ultimate quantity of interest is the internal alignment of the field border with the patient’s anatomy. This is generally verified with a port film image at some point. This work investigates the continuous monitoring of this quantity using the MV cine functionality available with all modern linacs. Method To assess the feasibility of MV cine imaging as a tool to observe the extent of DIBH, MV cine images have been obtained for all fractions of 10 patients in our clinic in parallel to the use of the Varian RPM system which is the current standard of care. The treatment was not altered. The MV cine images were analysed with an in-house Matlab routine to assess the distance between the posterior field edge and the rib structures in each image (lung depth, LD) in comparison to the distance expected from the treatment planning system (TPS).

Australas Phys Eng Sci Med ([email protected]), ([email protected]). 5Department of Medical Imaging, Fiona Stanley Hospital, Perth, Western Australia ([email protected]). 6Anatomical Pathology, PathWest, QEII Medical Centre, Perth, Western Australia ([email protected])

Results Interactive analysis of the patient cine images proved the feasibility of this approach. Figures show a representative image frame (top) and the range of the measured LDs for all fractions of this patient on the bottom (maximum, 95 % percentile, 75 % percentile, median, 25 % percentile, 5 % percentile, and minimum). TPS LD is shown in red. Conclusion Observing the alignment of the field border with the patient’s anatomy using MV cine imaging allows for direct assessment of the quantity of interest without any additional imaging dose or equipment. The next step is to quantitatively monitor DIBH during treatment using MV cine imaging.

P017 Fat quantification in human livers: A comparative study between magnetic resonance imaging, magnetic resonance spectroscopy, quantitative image analysis and conventional histopathology methods Thomas Greig1, Leon Adams2, Michael House3, Mike Bynevelt4, Lincoln Codd4, Anne Winsor5, Bastiaan DeBoer6, Roger Price1, Janette Atkinson1 1 Department of Medical Technology and Physics, Sir Charles Gairdner Hospital, Perth, Western Australia ([email protected]), ([email protected]), ([email protected]). 2School of Medicine and Pharmacology, Faculty of Medicine, Dentistry and Health Sciences, University of Western Australia, Perth, Western Australia ([email protected]). 3School of Physics, Faculty of Life and Physical Sciences, University of Western Australia, Perth, Western Australia ([email protected]). 4Department of Radiology, Sir Charles Gairdner Hospital, Perth, Western Australia

Introduction Hepatic steatosis is one of the most prevalent chronic liver diseases in Western countries and so the easier it is to accurately quantify fat accumulation within hepatocytes, the more likely the morbidity of hepatic steatosis can be reduced. This work compares the current gold standard for the assessment of liver fat (histopathology assessment by a pathologist) with image morphometry of biopsy samples and non-invasive techniques such as, three-point Dixon magnetic resonance imaging (3PD MRI) and magnetic resonance spectroscopy (MRS). The main objective of this comparison was to observe the correlation and relationship between the various quantification techniques in order to determine the viability of routine clinical implementation of the non-invasive quantification methods for the diagnosis/grading of hepatic steatosis. Method 30 patients that were diagnosed with hepatic steatosis and that were undergoing liver biopsy assessment as part of their routine clinical treatment were recruited. These patients were sent for MRI and MRS on the morning before their biopsy. A multi-echo in-phase/ out-phase image sequence was used for 3PD MRI quantification and a point-resolved spectroscopy sequence was used for MRS quantification. Biopsy samples were visually assessed by a reporting pathologist and were also digitally scanned for image morphometry analysis. Results and Conclusions The 3PD MRI and MRS estimations of hepatic fat fraction show a very strong correlation/agreement with the more objective image morphometry assessment of the biopsy slides (R2 [ 0.9). However, in comparison to the gold standard it was found that all of these methods significantly ‘underestimated’ the hepatic fat fractions by a factor greater than 3. These preliminary results are promising in that there is strong agreement between the image morphometry and the non-invasive quantification techniques, however, the sensitivity and specificity of these methods when compared to the gold standard needs to be investigated further to determine whether or not non-invasive techniques can be implemented clinically.

P018 Determination of spatial convolution kernels of a farmer chamber to evaluate the volume-averaging effect in a flattening filter-free beam F. Nelli1, J. Harwood1 1

Andrew Love Cancer Centre, University Hospital, Barwon Health, Geelong, Victoria, Australia. ([email protected]), ([email protected]) Introduction Flattening filter-free (FFF) beams have dose profiles with significant variation over distances commensurable to the dimensions of a Farmer chamber. This investigation presents a method for determining the spatial response of a Farmer chamber in order to calculate the volume-averaging effect in FFF beams under absorbed-dose calibration conditions. Method and Results In this study we implement a method to experimentally determine the spatial convolution kernel of a detector as presented by Garcia-Vicente et al. (1998). It is stated that the spatial convolution kernel of a detector is the derivative of the detector-measured profile of a step function with respect to detector position. In-air 9 and 12 MeV electron beams collimated by a lead shield in air were used as step dose distributions. The in-air profiles

123

Australas Phys Eng Sci Med Table 1 Volume averaging coefficient for a Farmer chamber on a 10 9 10 cm2 field size 10 MV FFF photon beam at different depths (SSD = 100 cm) Depth (mm)

23

50

100

200

300

Vol. avg. coeff.

0.992

0.992

0.993

0.994

0.994

were scanned with a FC-65P Farmer chamber and the spatial convolution kernels were calculated. The validity of the kernels was verified by measuring in-water dose profiles. Diode-measured dose profiles presenting negligible volumeaveraging effect were convolved with the Farmer-kernels. Comparison of these convolved profiles with Farmer-measured dose profiles verified the accuracy of the Farmer-kernels. A volume-averaging coefficient representing the effect of the Farmer chamber response on a 10 MV FFF photon beam was determined by convolving diodemeasured dose profiles with the Farmer-kernels. These results are presented in Table 1. The value corresponding to 10 cm depth is in agreement with published data (Sudhyadhom, 2013). Conclusion The method presented by Garcia-Vicente et al. for experimentally determining the spatial response of a detector was successfully applied to a Farmer ionization chamber. Subsequently, the chamber’s convolution kernels were used to calculate the volumeaveraging effect produced in a 10MV FFF photon beam. Reference 1. F. Garcı´a-Vicente, J.M. Delgado, and C. Peraza, Med. Phys. 25(2), 202 (1998). 2. A. Sudhyadhom, N. Kirby, C. Chuang, Med. Phys. 40(6) 491 (2013).

P019 Commissioning of an electron beam model in the Pinnacle treatment planning system and its validation across 4 Elekta linear accelerators B. Beeksma1, P. Vial1, J. Begg1, D. Truant1, J. Hellyer1, G. Gooze`e1 1

Liverpool and Macarthur Cancer Therapy Centres and Ingham Institute, Liverpool Hospital, NSW, Sydney, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]) Introduction Manually planned electron treatments typically do not include effects of tissue heterogeneity, irregular skin surfaces nor provides information about isodose distributions and anatomical coverage. Conversely, computer planned electron treatments such as a modified version of the Hogstrom electron pencil-beam algorithm used by the Pinnacle3 (Version 9.8, Philips Healthcare, USA) treatment planning system (TPS), does provide such information. Using a single TPS beam model for multiple linear accelerators simplifies work flow but requires precise beam matching and accurate dosimetric verification. Method An electron beam model was completed for electron energies of 6, 8, 10, 12, 15 MeV for an Elekta Synergy linear accelerator in the Pinnacle3 TPS. Based on methodologies and tolerances stated by IAEA TRS430, verification of the model was carried out by evaluating accuracy of TPS dose distributions against measured data for

123

Table 1 Mean Gamma pass rates for PDDs and profiles at various depths PDD

R100

R80

R50

Regular Fields 100SSD, 2 %/2 mm

96.7

97.4

97.4

86.5

Irregular Fields 100SSD 3 %/3 mm

97.0

100

97.8

89.6

Regular Fields 115SSD, 3 %/3 mm

99.6

100

98.7

86.9

Oblique Incidence, 100SSD 3 %/3 mm

100

95.8

98.6

84.5

regular field shapes, regular fields at extended source to surface distance (SSD), irregular fields, beams at an oblique incidence, complex surfaces and conditions containing heterogeneous material. Evaluation of model agreement across four linacs (two Synergy and two Versa HD linacs, all with Agility multileaf collimators) was then carried out by assessment of beam characteristics and cut-out factors. Results Table 1 summarizes the gamma criteria pass rates for computed and measured beam characteristics for profile penumbra and PDDs for all electron energies and field sizes used in the verification of the model. For all machines, across all energies, the maximum variation in measured vs TPS data for R100, R80 and R50 was 3.0, 2.0 and 0.5 mm respectively. Conclusion A single electron beam model was created within the Pinnacle TPS and successfully validated across four Elekta linacs. References 1. Al-Ghazi, M., Sehgal, V., Sanford, R and Chung, H (2007), ‘‘Experimental investigation of the implementation of Fermi-EygesHogstrom electron beam model of the Pinnacle3 system at extended SSDs,’’ Medical Dosimetry, vol. 32, no. 3, pp. 200–203 2. International Atomic Energy Agency (2004), Techanical Report Series No.430 Commissioning and quality assurnace of computerized planning systems for radiaiton treatment of cancer. Vienna: International Atomic Rnergy Agency

P020 An introduction to flattening filter free data collection, TPS modelling and VMAT QA D. P. Truant1, A. Gray1, B. Beeksma1, J. Hellyer1, R. Short1, P. Vial1, S. Rajapakse1, V. Nelson1, A. Ceylan1 1 Cancer Therapy Centre, Liverpool and Campbelltown Hospital, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected])

Introduction Flattening filter free (FFF) photon beams are a relatively new addition to the treatment techniques available with a conventional linear accelerator. Two Elekta Versa HD linear accelerators were installed during the 12 month period from September 2014 both including 6 MV and 10 MV FFF. This work summarises our recent experience in the commissioning, beam matching and modelling of two linear accelerators with a single treatment planning system (TPS) beam model for FFF VMAT SBRT applications. Method Standard data collection was carried out during acceptance for beam matching purposes. Data collection for Pinnacle V9.6 TPS

Australas Phys Eng Sci Med modelling purposes involved the use of diamond and diode detectors and GafChromic EBT3 film to manage the small field sizes, and improved accuracy for the profiles and penumbra. The model was generated to match field sizes only up to 20 9 20 cm, given the clinical focus of FFF treatments will be for relatively small tumours. Initial model verification followed the IAEA TRS430 methodology. Pre-clinical patient measurements of prostate, lung, and liver SBRT plans used a variety of dosimeters (ionisation chambers, diodes, film, EPI) and phantoms (static and dynamic) to verify accuracy. Results Beam matching was achieved between machines within 1 % dose difference within open fields and 2 mm distance to agreement in penumbra. Additionally, the energy of the flattened beam was matched to with 1 % of the FFF beam. The data imported into the TPS was solely collected using a diamond detector. Initial patient dosimetric results met the same criteria as set for flat beam hyprofractionated cases. Conclusion Two FFF energies were successfully beam matched across two linear accelerators, and modelled with a single TPS model for each energy. TPS modelling and verification followed the same procedure as for flattened beams with similar resulting accuracy.

P021 Validation of an elekta beam modulator MonteCarlo model Sri Herwiningsih1,2, Andrew Fielding2 1

Physics Department, Brawijaya University, Malang, Indonesia. Science and Engineering Faculty, Queensland University of Technology, Brisbane, Australia. ([email protected]), ([email protected])

2

Introduction The Elekta Beam Modulator collimator system, with a leaf width of 4 mm, was designed to facilitate the precise delivery of small field stereotactic treatments. This work aims to develop and validate a BEAMnrc Monte-Carlo model of the Elekta Axesse linear accelerator equipped with a Beam Modulator collimation system. Method The BEAMnrc Monte-Carlo model was pre-commissioned for large fields against measured dosimetry data. Electron energy and the width of the electron beam were optimised during commissioning process. The optimised parameters were then fine-tuned for small fields by simulating 6 MV photon beam for square fields of 1.6, 2.4 and 3.2 cm2 in a water phantom. Simulated lateral profiles (at depths of 1.5, 5 and 10 cm depths) were compared with the measured profiles. Measured and simulated output factors for different field sizes were also compared during as part of the commissioning process. Results The commissioning process found a best match for measured and simulated dosimetry data for an incident electron energy of 6.2 MeV and an elliptical Gaussian electron beam characterised by a FWHMx of 0.2 cm and FWHMy of 0.3 cm. An agreement of 1 % for the dose within the field was achieved between the simulated and the measured profiles. The exception was for the inline profile of the 1.6 cm 9 1.6 cm2 field at a depth dmax that was found to have a 1.5 % dose difference. The distance-to-agreement within the penumbra region for the measured and simulated lateral profiles was better than 1 mm. Measured and simulated output factors were found to agree to within 1 %. Conclusion The BEAMnrc Monte-Carlo model of the Elekta Beam Modulator has been validated for field sizes down to 1.6 cm and shows suitability for studying clinical stereotactic treatments. Ongoing study employs the model for lung SBRT plan verification.

P022 Case Study: The patient swallowed what? Did you just say a smoke detector? K. Hickson1, P. Collins2, D. Badger1, S. Toomey3 1 SA Medical Imaging, The Queen Elisabeth Hospital, Woodville South, SA, Australia ([email protected]), ([email protected]). 2SA Medical Imaging, Royal Adelaide Hospital, Adelaide, SA, Australia ([email protected]). 3SGS Australian Radiation Services, Blackburn North, VIC, Australia ([email protected])

Introduction A 30 year old female patient presented to TQEH Emergency Department after the patient dismantled a domestic Ionisation Chamber Smoke Detector (ICSD) and swallowed the source in an apparent suicide attempt. Domestic ICSDs almost exclusively use 241 Am with an activity of approximately 37 kBq. In most cases the radioactive source is encapsulated onto a very small foil disc approximately 3–5 mm in diameter. The source itself is very secure and access is only possible through a deliberate act. This Case Study reports the possible effective dose to the patient. Method A dosimetry calculation was made based on information reported by Rundo et al. (1977), who documented the ingestion of two 241Am source foils. Rundo et al. found that the foils remained in the GI tract for 16 and 24 days; after passing the foils only 1 % of the original activity was lost to the patient. Results The 20 mSv Annual Limit of intake via ingestion for 241Am is 100 kBq (Delacroix et al., 2002). In this case however the total free activity would be in the order of 370 Bq (1 % of the original activity) as most will remain on the foil. The committed effective dose per unit intake is reported as 2.0 9 10-7 Sv.Bq-1 (Delacroix et al., 2002). The calculated effective dose was therefore found to be *74 lSv. The patient had a bowel washout and was assessed in Emergency as impulsive rather than suicidal. Conclusion This incident occurred due to the mental state of the patient. Although this type of incident is rare it is difficult to implement measures to reduce the likelihood of a similar incident from occurring in the future. Based on a retrospective dosimetry calculation the radiation dose received by the patient is in the order of 2 chest X-rays. References 1. Rundo, J., W. D. Fairman, M. Essling, and P. R. Huff. (1977). Ingestion of Am-241 sources intended for domestic smoke detectors: report of a case. Health Phys. 33(6):561–566. 2. Delacroix, D., Guerre, J. P., Leblanc, P., Hickman, C. (2002) Radionuclide and radiation protection data handbook, Rad. Prot. Dos. 98:1.

P023 Monte Carlo simulation of kQclin,Qmsr for small field detectors and the Elekta Agility collimation system P. H. Charles1,2, S. Ibrahim3, D. Paynter4, D. I. Thwaites5 1 Princess Alexandra Hospital, Brisbane, Australia. 2Science and Engineering Faculty, Queensland University of Technology Brisbane ([email protected]). 3Princess Alexandra Hospital, Brisbane, Australia ([email protected]). St James University Hospital, Leeds, UK. University of Sydney, Australia

123

Australas Phys Eng Sci Med Introduction To calculate small field output factors an additional sensitivity correction factor (kQclin,Qmsr) needs to be applied, when the density of the detector is different to water1,2. This factor depends on field size, detector, energy and linac. Monte Carlo modelling can accurately calculate these factors; however no studies have done this for the Elekta Agility. This work fills this gap by using Monte Carlo simulations to calculate kQclin,Qmsr for the Elekta Agility for several small field detectors. Method The detectors used included: PTW 60017 electron diode, PTW 60016 photon diode, Sun Nuclear EDGE detector and Exradin W1 scintillator. Output ratios were simulated for square fields with side lengths 5–30 mm (SSD = 95 cm, depth = 5 cm in water, reference field size = 30 mm). Output ratios were simulated for each detector, and a detector-less geometry. kQclin,Qmsr for each detector/field size combination was calculated by dividing the output ratio for the detector-less geometry by the output ratio for the detector. Results kQclin,Qmsr for the PTW electron diode were 0.936, 0.970 and 1.000 at field sizes of 5, 10 and 20 mm respectively. These values were 2.6, 1.0 and 0.0 % lower than those previously experimentally measured3. However these results agreed closely (within 1 %) to other studies that use Monte Carlo simulations of different machines (e.g. Varian iX and the Elekta Synergy); as did the results for the other detectors. Conclusion kQclin,Qmsr has been simulated for various detector/field size combinations for the Elekta Agility collimation system. The results from this study are similar to other Monte Carlo calculated results from different machines with similar energies. This lends weight to the notion of universally applied kQclin,Qmsr factors. References 1. Alfonso, R. et al. (2008) A new formalism for reference dosimetry of small and nonstandard fields. Med. Phys. 35 pp 5179 2. Scott, A. et al. (2012) Characterizing the influence of detector density on dosimeter response in non-equilibrium small photon fields. Phys. Med. Biol. 57 pp 4461. 3. Cranmer-Sargison, G. et al. (2013) Small field dosimetric characterization of a new 160-leaf MLC, Phys. Med. Biol. 58 pp 7343.

P024 Sensitivity of a liquid filled ionization chamber array for detecting MLC errors in stereotactic radiotherapy plans P. T. O’Connor1, V. Seshadri1, C. E. Jones1, P. H. Charles1,2 1

Princess Alexandra Hospital, Brisbane, QLD, Australia ([email protected]). 2Science and Engineering Faculty, Queensland University of Technology, Brisbane, Australia Introduction The Octavius 1000 SRS array (PTW, Frieburg, Germany) was investigated to evaluate if the device had the high sensitivity and resolution required for effective quality assurance of stereotactic radiotherapy treatment plans. Method The sensitivity of the Octavius 1000 SRS (PTW, Frieburg, Germany) for detecting small (sub-millimetre) multi-leaf collimator (MLC) alignment errors in static square fields (side length 15–40 mm) and clinical stereotactic radiotherapy conformal arc fields was investigated. The commonly used gamma pass rate metric was used to compare detector array measurements to the dose calculated with the BrainLABTM treatment planning system. Results The detector array exhibited a drop in pass rate between fields without error and those which had MLC errors introduced. The gamma pass rates were evaluated as a function of MLC position error (MLC error size 0.1–2.5 mm). Drops in gamma pass rate increased as

123

MLC positional error increased, up to a maximum decrease of 4.5 % (gamma criteria 3 %, 1 mm) for a 0.8 mm error introduced to 15 mm square field. Conclusion This study showed that the Octavius 1000 SRS array could be a useful tool for applications requiring the detection of small geometric delivery uncertainties.

P025 Adaptive radiotherapy for bladder cancer patients using empty and full bladder imaging P. Juneja1,2,3, P. Hunt1, H. Caine1, J. Booth1,2, D. Thwaites2, A. Kneebone1, T. Eade1 1

Northern Sydney Cancer Centre, Royal North Shore Hospital Sydney, Australia. 2School of Physics, University of Sydney, Australia. 3Kolling Institute of Medical Research, Royal North Shore Hospital Sydney, Australia ([email protected])

Introduction The bladder can undergo large day-to-day deformations over the course of radiotherapy and therefore generous margins (up to 20 mm) are utilised for bladder planning target volumes (PTVs). A common goal of various bladder adaptive radiotherapy (ART) methods is to reduce irradiation of normal tissue while maintaining target coverage. The goal of this retrospective study is to develop and evaluate bladder ART, based on information on empty and full bladder volumes, applicable from the first day of treatment. Method Deformations between empty and full bladder were used to construct bladder anisotropic-PTVs (a-PTVs) through two methods, deformable image registration (a-PTVDIR) and interpolation (aPTVINTERP). For each patient and method, four a-PTVs were constructed such that a-PTV1 was the largest and a-PTV4 was the smallest and the a-PTVs covered at least the bladder volume plus 5 mm margin. Five previously treated patients (a total of 100 fractions) were used to develop and evaluate a-PTVDIR and a-PTVINTERP. These a-PTVs were compared to the current clinical standard (convPTV) of 10 mm uniform margins. Results The a-PTVs based on the two methods have been successfully constructed. Evaluation of the a-PTVDIR found that the smaller a-PTVs, i.e. a-PTV4DIR and a-PTV3DIR were appropriate in 87 % of the fractions, while a-PTV2DIR and a-PTV1DIR were required in 12 % of the fractions respectively. The use of the a-PTVDIR reduced the PTV volume by 32 % (28–36 %) as compared to conv-PTV. Conclusion The preliminary results indicate that the use of a-PTVDIR could result in a substantial decrease in the course averaged planning target volume. This reduction in the PTV is likely to decrease the radiotoxicity. The evaluation of a-PTVINTERP is currently underway.

P026 Clinical deliverability of modulated radiotherapy treatments for anal carcinoma T. Kairn1, J. Anderson1, S. B. Crowe2, M. Lah1 1

Genesis Cancer Care Queensland, Brisbane, Australia ([email protected]), ([email protected]), ([email protected]). 2Cancer Care Services, Royal Brisbane and Women’s Hospital, Brisbane, Australia ([email protected]) Introduction Although modulated radiotherapy techniques (IMRT and VMAT) have been shown to produce planned dose distributions with conformal high-dose regions that cover the anus and pelvic

Australas Phys Eng Sci Med nodes and avoid nearby radiosensitive organs, the reliability with which these planned dose distributions can be delivered requires examination. This study provides a detailed evaluation of the deliverability of radiotherapy treatment plans for anal carcinoma, in comparison with the simpler modulated treatment plans that are used for treating prostate carcinoma. Method This study evaluated 10 modulated radiotherapy treatment plans for anal carcinoma and compared the results with data from 70 prostate radiotherapy treatment plans. The anal carcinoma plans were designed to treat large volumes which encompassed pelvic nodes. All treatment plans were evaluated using established modulation indices, complexity scores and small-aperture scores. Pass rates from routine quality assurance (QA) checks were used to quantify the treatment deliverability. Results Overall mean QA pass rates were 5 % lower for the anus plans than for the prostate plans, with 10 % of the anus treatment beams and 0 % of the prostate treatment beams failing their QA tests. The anus treatment plans produced higher modulation indices and complexity scores than the prostate plans. Elevated small aperture scores, correlated with reduced QA pass rates, were observed in 65 % of anus treatment plans and only 30 % of prostate treatment plans, despite the larger target volumes treated by the anus plans. Conclusion These results suggest that the QA pass rates for anal carcinoma treatment plans may be detrimentally affected the use of small beam segments to produce varying radiation intensities across relatively large beam areas. The accurate delivery of planned treatment doses may therefore be improved by limiting the number of segments per beam, or increasing the minimum allowed MLC leaf aperture size, when optimising treatment plans for large volumes.

P027 Validity of conformity indices for categorising radiotherapy treatments for anal and rectal carcinoma T. Kairn1, J. Anderson1, S. B. Crowe2, M. Lah1

for conformality may be too prescriptive for these anus and rectum treatment plans; only 50 % of 3DCRT plans, 67 % of IMRT plans and 93 % of VMAT plans can be called conformal, under this definition, despite the isodoses and DVHs for all plans achieving clinically acceptable organ-at-risk sparing. Conclusion Conformity indices have the potential to be valuable tools for the bulk assessment of treatment plan quality, however the physical interpretations and limitations of all proposed indices should be thoroughly evaluated before such use is adopted.

P028 Six isocentres and a piece of film: Comprehensive end-to-end testing of a cranial stereotactic radiosurgery system T. Kairn1,2, M. West1 1

Genesis Cancer Care Queensland, Brisbane, Australia ([email protected]). 2Queensland University of Technology, Brisbane, Australia ([email protected]) Introduction End-to-end testing, which usually includes either dosimetric or geometric testing (Murphy & Cox 1996, Ramakrishna et al. 2010), is a fundamental part of contemporary radiotherapy quality assurance. This work provides a simple method for comprehensive end-to-end testing measurements for a stereotactic radiosurgery system. Method A head phantom was immobilized using a thermoplastic mask and CT scanned within a stereotactic localizer box, for treatment as a stereotactic radiosurgery case. A six isocentre, seven field, non-coplanar radiotherapy treatment was planned and delivered to the phantom, using a beam arrangement that allowed a range of important geometric and dosimetric measurements to be made using one sheet of film placed transversely through the phantom.

1

Genesis Cancer Care Queensland, Brisbane, Australia ([email protected]), ([email protected]), ([email protected]). 2Cancer Care Services, Royal Brisbane and Women’s Hospital, Brisbane, Australia ([email protected]) Introduction Simple dose-volume metrics, termed ‘conformity indices’, may be valuable in the emerging area of bulk treatment plan audit and review, enabling the use of a single parameter to categorise the dosimetric quality of hundreds of treatment plans. This study evaluates several recognised conformity indices in terms of their suitability as surrogates for more complex radiotherapy treatment plan parameters. Method Six intensity modulated radiotherapy (IMRT) plans, five volumetric modulated arc therapy (VMAT) plans and fifteen 3D conformal radiotherapy (3DCRT) treatment plans for anal and rectal carcinoma were evaluated. Five types of conformity index were evaluated and compared with standard isodose and dose volume histogram (DVH) results, for each plan. Results RTOG conformity indices and healthy tissue overdosage factors for the 3DCRT plans are dominated by the relatively large volumes of healthy tissue that receive the prescription dose during treatments of small targets. RTOG conformity indices for the IMRT and VMAT plans are closer to unity. The treatment volume coverage and underdosage factors reliably describe the homogeneity of dose to the target volume, with the IMRT and VMAT doses being slightly (but not significantly) more heterogeneous than the 3DCRT plans. The use of a healthy tissue conformity index of 0.6 as the threshold

Results The treatment plan used in this study permitted the efficient dosimetric evaluation of the effects of treating with very small fields (2 % difference between planned and delivered dose for a 5 9 6 mm2 field), static conformal arcs (3 % difference between planned and delivered dose for an 80 arc), and treating through a thick layer of bone (up to 4 % difference between planned and delivered dose, for a pencil-beam algorithm). Additionally, this treatment plan allowed the geometric accuracy of beam delivery to be straightforwardly verified. All measured positions of maxima in high-dose regions were found to agree within 1.00 ± 0.35 mm with the planned locations of the isocentres, in terms of their distances from physical landmarks in the phantom and their differences from each other. Conclusion Dosimetric end-to-end tests can be used to quantify dose calculation and delivery accuracy, for linac and treatment planning system commissioning and quality assurance, and geometric end-toend tests can be used to quantify the geometric accuracy of image guidance and treatment delivery. The method developed in this study can be used to fulfill both of these goals, within one treatment session.

123

Australas Phys Eng Sci Med References 1. Murphy, M. J., Cox, R. S. (1996) Medical Physics 23(12): 2043–2049 2. Ramakrishna, N. et al. (2010) Radiotherapy and Oncology 95: 109–115

P029 Commissioning of desktop scanners for use in radiochromic film readout T. Kairn1,2, B. Sutherland1 1 Genesis Cancer Care Queensland, Brisbane, Australia ([email protected]). 2Queensland University of Technology, Brisbane, Australia ([email protected])

Introduction As the use of very small or highly modulated beams for radiotherapy treatments has become more common, the benefits of acquiring high-resolution two-dimensional dose measurements using radiosensitive films has become apparent [1,2]. The achievable accuracy of a film dosimetry programme is substantially affected by the choice and use of the film readout system. This study investigated the performance and reliability of three desktop scanners, with a view to providing an indication of the extent to which scanning technology and scanning method may affect measurement results. Method An EPSON V700 flatbed scanner (purchased 2008) and two EPSON V800 flatbed scanners (purchased 2014) were each used to acquire a set of scans with samples of EBT3 radiochromic film in various positions on the scanning beds. These scans were used to investigate scanner warmup time, scan area limits and scan area homogeneity, and to quantify the sensitivity of dosimetry measurements to film position and orientation. Results The results from the three scanners were generally similar, although the older scanner was found to require 10 scans, after ‘‘warmup’’, to fully stabilise, while the two newer scanners produced stable results after fewer than 5 scans. All scanners were found to have maximum scanning areas that were in agreement with each other to within 2 mm. For all scanners, rotating the film on the scanning bed by 90o resulted in changes in the red-channel pixel value of approximately 5 %, and placement of calibration films at edges of the scanning bed, rather than at the centre, generally increased dosimetric uncertainty from 1–2 % to 6–8 %. Conclusion All three scanners were released for clinical use, subject to specific recommendations regarding the positioning and orientation of the films during scanning. This study provides an example of a useful commissioning procedure for evaluating new or existing film scanners. References 1. Kairn, T. et al. (2011) Australas. Phys. Eng. Sci. Med. 34(3): 333–343 2. Moylan, R. et al. (2013) Australas. Phys. Eng. Sci. Med. 36(3): 331–337

P030 Use of the triple channel dosimetry for EBT3 Gafchromic film in quality assurance of stereotactic radiotherapy plans A. Kazi1, G. Godwin1, E. Baveas1, D. Wallace1 1 Radiation Oncology Queensland, Gold Coast University Hospital ([email protected]), ([email protected]), ([email protected]), ([email protected])

123

Introduction As part of the stereotactic radiotherapy QA at Radiation Oncology Queensland Gold Coast, patient plans are delivered to Gafchromic EBT3 film in an IBA Dosimetry cube phantom. The films are scanned with an Epson 11000XL scanner, converted to dose in FilmQA Pro 2014 (ISP Advanced Materials) using triple channel dosimetry, and compared to the planned dose distribution using a gamma index analysis. It was noted that at doses about 14 Gy the gamma index results were lower than expected, but there was large improvement when single channel dosimetry was used. The aim of this study was to further investigate the reason for this effect. Method The EBT3 film was placed vertically in the water tank and irradiated with a 10x10 cm 6MV beam delivering about 14.5 Gy at 10 cm depth. Central axis depth dose curves were created using the triple channel and single channel dosimetry methods, and compared to departmental reference data. The scanned image pixel value with depth was also plotted. Results Analysis of the depth dose curves showed that there was a region where the triple channel method depth dose deviated from the reference depth dose curve, creating a discontinuity (a ‘‘step’’). This effect wasn’t present in depth doses created with a single channel method. Analysis of the single channel calibration curves showed that the gradients of the blue and green channel calibration curves are equal at about 14 Gy, indicating this may be possible reason for the break down in triple channel dosimetry. The curve representing pixel value with depth showed no discontinuity. Conclusion There appears to be a dose region where the triple channel dosimetry may not be able to be applied and single channel dosimetry using the channel with the largest gradient in calibration curve for this dose region could be used. The results are under further investigation.

P031 Spectroscopic dosimetry for eye plaque brachytherapy QA A. M. Kejda1, D. L. Cutajar1, M. R. Weaver1, A. B. Rosenfeld1 1 Centre for Medical Radiation Physics, University of Wollongong, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected])

Introduction Spectroscopic dosimetry has been developed for LDR prostate brachytherapy dosimetry employing I-125 sources [1], providing a more sensitive measurement than direct current modes typically used in diode based detectors. The concept of spectroscopic dosimetry was expanded to eye plaque brachytherapy, where a prototype dosimetry system for fast eye plaque brachytherapy verification has been developed. This research outlines the design, testing and Monte Carlo optimisation of the current and future prototype design for the eye plaque QA system. Method The spectroscopic QA system consists of a 15 cm probe; containing a diode and on-board pre-amplifier system at the tip, and a readout system; housing the amplifier and dose calculation circuitry. The latter counts events corresponding to photons above 20 keV, approximating the dose rate. The positioning system uses a linear stage and stand allowing 0.5 mm increment depth dose measurements, to 1 mm from the plaque surface. The experimental investigation included depth dose measurements using the spectroscopic QA system and ROPES eye plaque loaded with ten I-125 seeds in a water phantom. Monte Carlo simulations were performed using GEANT4 to model a full ROPES applicator loaded with 10 I-125 seeds, to study the accuracy of spectroscopic dosimetry at close range to the plaque, as well as to optimise encapsulation designs for the probe.

Australas Phys Eng Sci Med Results Experimental results showed a close comparison to the TG43U1 calculation and Geant4 model indicating a functioning eye probe. Conclusion The spectroscopic QA system has proven a legitimate tool for use in eye plaque QA, providing fast dosimetric measurements of I-125 based eye plaques prior to treatment.

used. Nine hundred questionnaires were distributed. IBM SPSS version 20 was used for statistical analysis. Results 752 questionnaires were filled and returned (83.6 %). Questionnaires with missing answers were excluded. The respondents were nurses (64.3 %), (29 %) were physicians of different level of education; the remaining (6 %) are technicians and other staff. It included questions differentiating between ionising and non-ionising radiation, and the organ sensitivity. The average score is 37.6 ± 25.0 in (significant at p 0.05). Nurses scored lowest compared to all. On the other hand, senior house officers scored highest among all, including consultants in both level was very low in nurses compared to. Conclusion The results from our questionnaire indicated the inadequacy of awareness of ionising radiation. For the rapidly developing field of radiation imaging, continuous updated knowledge is required to assure no patient is exposed to unnecessary radiation and subsequent risk of developing cancer. Formal training is should be mandatory to raise the awareness and referral practice in imaging. References

Fig. 1 The QA system. A. detector probe, B. Motorised gantry, C Eye plaque phantom, D. Movement control unit, E. Eye plaque phantom insert References 1. Cutajar DL, Takacs GJ, Lerch MLF, et al., (2006) Intraoperative solid-state based urethral dosimetry in low dose rate prostate brachytherapy, IEEE Transactions on Nuclear Science, Volume 53, Number 3, pp 1408–1412

P032 Health professional awareness on radiation protection in Saudi Arabia Mawya A. Khafaji1, Sarah K. Hagi1, Abeer A. Albar2 1

Department of Radiology t, King Abdulaziz University Hospital. KSA ([email protected]), ([email protected]). 2School of medicine, King Abdulaziz university hospital. KSA ([email protected]) Introduction We are exposed to ionizing radiation on daily basis, either from background or technologybased. Diagnostic radiographs are the most common source of technology-based radiation exposure, and contribute 50 % of the annual allowed dose received by an individual. There are increasing data about risk of cancer associated with ionizing radiation used for medical purposes and that it has dose dependent effect on the human body. Physicians are not accurately aware about radiation dose received by their patients 1, 2. Some tend to underestimate the radiation doses and others think it is not an important factor when choosing which modality to order. Methodology A cohort study was conducted on health professionals of all levels at King Abdulaziz University hospital. A modified version of a validated questionnaire by Reichmuth3 in 2013 in Malta was

A survey of awareness of radiation dose among health professionals in Northern Ireland. Soye JA, Paterson A. Br J Radiol. 2008 Sep; 81(969):725–9. Radiation protection awareness in non-radiologists. A D Quinn, C G Taylor, T Sabharwal, and T Sikdar, BJR Published Online: January 28, 2014, Volume 70, Issue 829 Radiation awareness among clinicians in Malta, L. Reichmuth, K. Micallef, A. Mizzi; Tal-Qroqq, Sliema, ATTARD, Poster ECR 2013.

P033 Investigation of cine image header gantry angles and their effect on a 3D EPID dosimetry method K. Legge1, P. Greer1,2, J. Martin2, J. O’Connor1 1 School of Mathematical and Physical Sciences, University of Newcastle, NSW, Australia ([email protected]). 2 Calvary Mater Hospital, Newcastle, NSW, Australia ([email protected]), ([email protected]), ([email protected])

Introduction Cine images acquired using Varian’s Eclipse software save incorrect gantry angles in the header. This work investigates the discrepancy between the header angles and the true gantry angle and to quantify the impact of these incorrect gantry angles on a threedimensional (3D) VMAT dosimetry technique which uses these cine images and allows remote credentialling for multicentre studies. Method The EPID dosimetry method derives a 3D composite dose distribution in a virtual water-equivalent cylindrical phantom from EPID images measured in air. It corrects for backscatter artefacts from the EPID support arm, converts to fluence and then calculates dose in the phantom. The 3D dose is compared to a treatment planning system verification plan calculated on the phantom. A VMAT plan was delivered and cine images acquired with differing frame averages using the on-board imaging software. Individual frames were acquired simultaneously using a research framegrabber system. The 3D dose was constructed from the cine images and also the framegrabber images and both doses compared to the treatment planning system dose. The header gantry angles in the cine images were compared to the gantry angles recorded using the framegrabber system.

123

Australas Phys Eng Sci Med Results A smaller number of frame averages in the cine images yielded a better match between the cine and framegrabber 3D dose results and a closer match between the header and framegrabber gantry angles. Header gantry angles in the cine images varied up to 4.5 degrees from those recorded by the research framegrabber. Conclusion The cine image header angle is closer to the true gantry angle when a smaller number of frames is being averaged. A quality assurance method has been developed for use on VMAT plans using 3D EPID dosimetry. The technique used here will be useful for comparison of plans between centres in multicentre clinical trials.

P034 Kilovoltage intrafraction monitoring of prostate SBRT boost treatments K. Legge1, P. Greer1,2, J. Ng3, J. Booth4, P. Keall3, J. Martin4, 1 J. O’Connor 1

School of Mathematical and Physical Sciences, University of Newcastle, NSW, Australia ([email protected]), ([email protected]). 2Calvary Mater Hospital, Newcastle, NSW, Australia ([email protected]). 3 Institute of Medical Physics, University of Sydney, NSW, Australia ([email protected]), ([email protected]). 4Royal North Shore Hospital, NSW, Australia ([email protected]). 4 Calvary Mater Hospital, Newcastle, NSW, Australia ([email protected]) Introduction The PROMETHEUS trial commenced in Australia last year and provides a prostate SBRT boost with a Rectafix rectal sparing device in place. Accuracy of Kilovoltage Intrafraction Monitoring (KIM) software was tested using static and moving phantoms and the dose delivered was measured with an ion chamber. Images were taken during patient treatments and analysed using KIM software to assess the effectiveness of the Rectafix in preventing prostate motion. Method Two-dimensional kilovoltage images of fiducial markers in a phantom were taken while a VMAT SBRT plan was delivered. The phantom was offset in three planes to test static accuracy, and a programmable motion phantom recreated six real prostate trajectories to test moving accuracy. The imaging dose during an average treatment was calculated from ion chamber measurements. Images were also acquired during patient treatments. Fiducial marker positions were manually segmented in the initial two-dimensional projection, then KIM software was used to findsthe most probable marker positions in three dimensions. KIM software was used post-delivery to analyse prostate motion in three dimensions for each image set. Results The KIM software measured static shifts to within ±0.7 mm in all three planes. Four of the six motion trajectories passed a criteria of \1.0 mm mean difference and \1.0 mm standard deviation between the KIM trajectory and original trajectory. Two cases failed due to a higher standard deviation value, potentially due to very high frequency motion. Motion over 2 mm in any direction was very rare during patient treatments. The imaging dose over two SBRT fractions was calculated to be 157 mGy. Conclusion Prostate motion was usually less than 2 mm during patient treatments, suggesting that the Rectafix is useful in reducing prostate motion. Use of KIM software online during treatment would be valuable for monitoring prostate position during high dose SBRT treatments.

123

P035 Real time in vivo MOSFET dosimetry for prostate SBRT boost treatments K. Legge1, P. Greer1,2, D. Cutajar3, A. Wilfert2, J. Martin2, A. Rozenfeld3, J. O’Connor1 1

School of Mathematical and Physical Sciences, University of Newcastle, NSW, Australia ([email protected]), ([email protected]). 2Calvary Mater Hospital, Newcastle, NSW, Australia ([email protected]), ([email protected]), ([email protected]). 3Centre for Medical Radiation Physics, University of Wollongong, NSW, Australia ([email protected]), ([email protected]) Introduction The PROMETHEUS trial commenced in Australia last year and provides a prostate SBRT boost consisting of two fractions of 9.5 Gy each with a Rectafix rectal sparing device in place. In vivo dosimetry of the rectal wall using MOSFETs attached to the Rectafix was performed as a phantom study and for patients enrolled in the trial. Method A dual MOSkin detector was attached to the anterior surface of the Rectafix device and inserted into an anthropomorphic phantom. MOSFET voltages were read out at 1 Hz during delivery and converted to dose using a calibration factor. Dynalog files were acquired during treatment so that the position of the gantry at each point in time was known. The dose at the MOSFET location was extracted from the treatment planning system in 5 degree increments for each delivered arc. The MOSFET dose and planning system dose over the course of each treatment arc were then compared. This process was repeated for patients enrolled in the PROMETHEUS trial. Results Measured dose was within 10 % of planning system dose for all deliveries monitored. Uncertainty in placement of the MOSFET detector within the patient and of the verification point in the planning system are major sources of discrepancy, as the detector is placed in a high dose gradient region during treatment. Conclusion This method can be used to verify treatment deliveries and Rectafix positioning within patients. Further developments could enable this method to be used during high dose treatments to monitor dose to the rectal wall to ensure it remains at safe levels.

P036 Sifting data from the clinical coalface: Implementation of a Radiation Oncology data mining system in Newcastle, Australia to aid clinical decisions S. Bhatia1, S. Walsh2, M. Field3, M. S. Barakat3, P. Greer4, J. Ludbrook4, A. Dekker2, L. Holloway5, S. K. Vinod6, D. Thwaites3, J. Lehmann1,3 1 Department of Radiation Oncology, Calvary Mater Newcastle, Australia ([email protected]). 2MAASTRO clinic, Maastricht, The Netherlands ([email protected]), ([email protected]). 3Institute of Medical Physics, The University of Sydney, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]). 4Calvary Mater Newcastle, Newcastle, Australia ([email protected]), ([email protected]). 5Department of Radiation Oncology and Ingham Institute, Liverpool Hospital, Australia ([email protected]). 6Department of Radiation Oncology, Liverpool Hospital, Australia ([email protected])

Australas Phys Eng Sci Med Introduction Recent applications for datamining in radiation therapy, notably from our collaborators in the MAASTRO Clinic and the NSW radiation oncology datamining network, develop predictive models and decision support systems (DSS). (Dekker et al. 2014) In the MAASTRO approach, using ‘distributed learning’, the model with its parameters travels between institutions while the patient data remains at each institution. Incomplete datasets reduce the number of patients available for inclusion in model building or testing. They also potentially introduce a bias in the model if data primarily from specific patient groups or radiation oncologists are missing. We report on our experience with NSCLC data selection and cleaning. Method Our institution has recently built a datamining system linked to the developing NSW datamining network, supported by MAASTRO and Sydney University collaborators. Clinical data sets were exported, anonymized and prepared for analysis with the MAASTRO DSS. Two year survival is determined relative to start of radiation treatment using the date of death (DoD). Data for the DSS’ five predictive factors are obtained: (1) Tumour volume is calculated with an in-house script deployed remotely to Eclipse (2) Forced Exhale Volume (FEV) and (3) ECOG status are found in free text search. (4) Gender is available in the database. (5) The number of positive lymph nodes is inferred from tumour stage. Results With 4285 initially retrieved patients, data preparation is in progress: Selection was made for NSCLC patients, stages I–III, and available DoD. Tumour volume calculation with a previously developed script was adapted for data mining; results are compared to those using the MAASTRO approach. Free text search for FEV and ECOG are ongoing. Conclusion Completing datasets for datamining is feasible and requires a team approach with sophisticated computer science components. Efforts should be made to prospectively keep all relevant patient data in a structured format. References 1. Dekker, A., S. Vinod, L. Holloway, C. Oberije, A. George, G. Goozee, G. P. Delaney, P. Lambin and D. Thwaites (2014). ‘‘Rapid learning in practice: a lung cancer survival decision support system in routine patient care data.’’ Radiother Oncol 113(1): 47–53.

P037 Constructing dosimetry problems using MrVoxel, GATE and GRNT E. McKay1 1

Dept. Nuclear Medicine, St. George Hospital, Sydney, Australia ([email protected])

Introduction Individual internal dosimetry calculations are of increasing interest for Nuclear Medicine practices with a therapeutic workload. The challenge is to find an optimal approach for a given problem, rather than simply applying a one size fits all solution. Software phantoms allow identical data to be viewed and analysed in different ways, providing a means to address this challenge. Aim To develop a tool for constructing internal dosimetry problems, comprising a collection of operator-defined projection images of a time-varying 3D dose distribution taken at operator-selected time intervals, along with their solutions in the form of 3D voxel dose distributions. Method The open-source MrVoxel software [1] provides tools for manipulating collections of planar and volume images and data tables via menu commands and Tcl scripts. Plug-in modules provide for data transfer between MrVoxel and the Monte Carlo codes GATE, for image simulation, and GRNT, for voxel-level dose deposition

calculations. Another plug-in simulates biokinetics by iterating a linear compartment model to produce time-activity curves (TACs). Results A problem is created by loading a voxel geometry and tabulating initial tissue activity and transfer coefficients. Tissue geometry is represented by collections of fractional voxels. Monte Carlo simulation is performed for each tissue, using GRNT to establish the distribution of dose around it and GATE to simulate its projections. The projections are scaled and aggregated using the TACs integrated over each nominated time-interval, with noise added via Poisson resampling. The resulting simulated images can be analysed using MrVoxel or exported in Interfile format. Dose distributions are likewise scaled using the TACs integrated over 10 decay half-lives, then aggregated to provide the problem’s solution. Conclusion The software has been developed and used to construct a simple 5-compartment dosimetry problem based on the ICRP 110 voxel phantom. References 1. McKay, E. (2003). A Software Tool for Specifying Voxel Models for Dosimetry Estimation. Cancer Biotherapy & Radiopharmaceuticals 18:3.

P038 An optically stimulated luminescence based 2D radiation dosimetry system J. W. McMahon1, G. V. M. Williams1, D. Thomas2 1

MacDiarmid Institute, Victoria University of Wellington, PO Box 600, Wellington 6140, New Zealand ([email protected]), ([email protected]). SCPS, Victoria University of Wellington, PO Box 600, Wellington 6140, New Zealand ([email protected]) Introduction New tissue-equivalent 2D dosimeters are of interest for verification of volumetric modulated arc therapy. General requirements include high spatial resolution, good reproducibility, and fast readout. For this reason we have been developing an optically stimulated luminescence (OSL) based 2D dosimeter system that uses fluorescent ion doped NaMgF3 [1–3]. OSL is a process where ionizing radiation creates trapped electrons and holes which are later caused to recombine by optical stimulation followed by the emission of photons. The integrated intensity of the OSL can then be related to the dose [1]. Method Polycrystalline NaMgF3:Eu3+ was made using a process described elsewhere [1]. It was ground into powder and mixed with a polymer to make 2D dosimeter sheets. The sheets were tested using 40 kV x-rays. They were read out using a 3D printer-based prototype reader and a 405 nm diode laser. Results 2D dosimeter sheets with thicknesses of up to 200 lm have been made with sizes of up to 10 9 10 cm. The sheets were exposed to approximately 2 Gy with 40 kV x-rays and read out using a prototype system that uses a focussed raster scanning laser mounted on a modified 3D printer. The emitted light was detected using photomultiplier tubes. A response was detected after irradiation and a spatial resolution of 10 line pairs per mm was measured. It was shown that the plates could be reused and the spatial inhomogeneity was measured. Previous measurements on NaMgF3:Eu2+ bulk crystals with a point dosimeter system have shown energy response comparable to an air filled ionisation chamber at 6 MV. Conclusion Our current results show that 10 9 10 cm 2D dosimeter sheets can be made with good spatial resolution. The next development phase involves testing at 6 MV, decreasing spatial inhomogeneity across sheets, and continued testing using larger sheet sizes.

123

Australas Phys Eng Sci Med References 1. Dotzler, C et al. (2007), Appl. Phys. Lett. 91:121910 (2007). 2. Williams, GVM et al., J. Lumin. 143:219 (2013). 3. Gaedtke, C et al. (2014), Radia. Meas. 71:258 (2014).

P039 Estimating radiation interaction coefficients from single energy CT S. M. Midgley1,2 1

William Buckland Radiotherapy Centre, Alfred Hospital, Melbourne, VIC 3181. 2South Australian Medical imaging, Flinders Medical Centre, Bedford park SA 5042 ([email protected]) Introduction Photon interaction coefficients are required for attenuation correction and absorbed dose calculations in Radiology, Nuclear Medicine and Radiotherapy. A method is investigated for obtaining these coefficients from CT Hounsfield numbers. Method The compositional dependence of atomic cross-sections is described using a polynomial in atomic number (Midgley 2004, 2011, 2014). The same model accommodates the attenuation coefficient and CT Hounsfield numbers (HN) for mixtures. The polynomial coefficients can be obtained by solving linear simultaneous equations formed by HN measurements with materials of known density and composition. Knowledge of atomic contribution to HN enable a virtual CT scan to be conducted and predict HN for tissues (ICRU 1989, 1992). Similar calculations using the tabulations (Cullen et al. 1989, Hubbell and Seltzer 1995) and mixture rule deliver attenuation coefficients and mass energy absorption coefficients for mono-energetic radiation 10 to 20 MeV. The two are combined to produce look up tables mapping HN to mono-energetic interaction coefficients. Results We consider measurements with a GE discovery (CT590) scanner, tissue substitute phantom (RMI-467) and other common materials. Published measurements with earlier generations of the tissue substitute phantom and different CT scanners are also considered (Constantinou et al. 1992, Schneider et al. 1996, Watanabe 1999, Schneider et al. 2000). Results are presented showing the relationship between predicted HN for tissues, electron density and photon interaction coefficients for healthy tissues and mono-energetic radiation. Conclusion A novel method is demonstrated for estimating photon interaction coefficients for tissues from CT HN.

Fig. 1 Atomic cross-sections divided by the attenuation coefficient for water, for the NIST tabulation (dotted lines) and for the CT scanner from measurements with RMI-467 tissue substitutes and other solid samples

123

Fig. 2 Scanner specific look up tables linking measured HN at 120 kVp for tissues to mono energetic linear attenuation coefficients

References Constantinou C Harrington J C and DeWerd L A (1992) An electron density calibration phantom for CT-based treatment planning computers Med. Phys. 21:325–7 Cullen D E et al. 1989 Tables and Graphs of Photon Interaction Cross-Sections from 10 to 100 GeV Derived from the LLNL Evaluated Photon Data Library (Springfield, VA: National Technical Information Service) Hubbell J H and Seltzer S M (1995) Tables of x-ray Mass Attenuation Coefficients 1 to 20 MeV for Elements Z = 1 to 92 and 48 Additional Substances of Dosimetric Interest (NISTIR 5632) (Gaithersburg, MD: National Institute of Standards and Technology) pp 1–79 ICRU 1989 ICRU Report 44 Tissue Substitutes in Radiation Dosimetry and Measurement (Bethesda, MD: ICRU) pp 37–8 ICRU 1992 ICRU Report 46 Photon, Electron, Proton and Neutron Interaction Data for Body Tissues (Bethesda, MD: ICRU) pp 5–16 Midgley SM (2004) A parameterisation scheme for the x-ray linear attenuation coefficient and energy absorption coefficient Phys. Med. Biol. 49:307–25 Midgley SM (2011) A model for multi-energy x-ray analysis Phys. Med. Biol. 56:2943–62 Midgley SM (2014) A method for estimating radiation interaction coefficients fro tissues from single energy CT Phys. Med. Biol. 59:7479–99 Schneider U, Pedroni E and Lomax A 1996 The calibration of CT Hounsfield units for radiotherapy treatment planning Phys. Med. Biol. 41:111–24 Schneider W, Bortfeld T and Schlegel W 2000 Correlation between CT numbers and tissue parameters needed for Monte Carlo simulations of clinical dose distributions Phys. Med. Biol. 45:459–78 Watanabe Y 1999 Derivation of linear attenuation coefficients from CT numbers for low energy photons Phys. Med. Biol. 44:2201–11 Keywords Computed tomography Hounsfield numbers, X-ray linear attenuation coefficient, mass energy absorption coefficient, electron density, attenauation correction, absorbed dose calculation (Libre office finds 296/300 words) Author Biographical details (42/50 words) The author has found employment with the non-destructive testing industry and clinical departments providing Nuclear Medicine, PET,

Australas Phys Eng Sci Med Radiology and Radiotherapy services. This presentation concerns research conducted outside of normal working duties, presently with the Department of Medical Imaging at Flinders Medical Centre.

P040 Evaluation of siemens’ metal artifact reduction in image space (MARIS) algorithm J. Miller1, T. Markwell1 1 Radiation Oncology Mater Centre, Metro South – Queensland Health, South Brisbane, QLD, Australia ([email protected]), ([email protected])

Introduction Implants encountered in radiotherapy patients commonly consist of titanium, steel or cobalt chromium alloy, with a density of 4–8.5 g/cm3. MARIS applies corrections for beam hardening in the projection and image domain to image projections containing metal to reduce artifact production1. Method A hip replacement (‘‘hip’’) was CT scanned with an adult pelvis protocol. Six scans were reconstructed, with the B31 s and five MARIS (MAR0 – MAR4) kernel strengths. Using the B31 s scan as a baseline, a difference image set was created for each MARIS reconstruction and analysed using profile comparison. The hip was also scanned with a 5MU (total) 6MV MegaVoltage Cone Beam CT (MVCBCT) at 5 MU/min2. The reconstructed images were analysed using ImageJ and the Relative Electron Density (RED) of the wax and hip estimated. Fusion of the CT scan with the MVCBCT allowed for manual correction of artifacts. Results and Discussion The difference images show MARIS acts within the hip and where void artifacts are present. Siemens suggest using the stronger MARIS kernels for high density implants. AAPM TG633 data was used to convert the hip’s RED to physical density, which resulted in a density of * 9.5 g/cm3. Analysis shows that whilst MAR4 overcorrects the areas of artifact, the MAR3 kernel shows greater delineation of structures in the scan and a reduction in the magnitude of the void artifact. Profiles through the difference image sets show the greatest CT number shift occurs within the hip using the MAR4 kernel. Conclusion With careful selection of MARIS kernel, void artifacts produced in the CT scan of hip replacement patients can be reduced. It is hoped that the release of the next generation algorithm will increase this benefit. Further work with MARIS is planned on other prostheses in conjunction with MVCBCT.

([email protected]), ([email protected]) Introduction Four-dimensional computed tomography (4DCT), a technological innovation in the field of CT to account for tumour motion, was introduced at Waikato Hospital. Similar to the introduction of any other new technologies, a series of tests such as accuracy of phase binning, shape, volume and CT number as applied to radiation treatment planning were evaluated [1]. Method A modified motion phantom supplied by Varian was scanned with retrospective technique using Aquilion LB (Toshiba Medical System Corporation). Datasets were binned into phases at 10 % interval. Phase binning accuracy was determined by identifying relative positions of the ball bearings (BBs) on a CT dataset, position of BBs on other phases was estimated and compared it with the actual phase. A dynamic thorax phantom was also scanned for a range of amplitudes and periods reflecting the clinical environment [2]. The resulting 4DCT dataset were binned by phases into 10 intervals and accuracy of shape, volume and CT number was assessed using software tools in the treatment planning system (TPS). The metric used to assess geometric distortion was the ratio of the y-dimension to the x-dimension. TPS software tools were used to calculate the volumes of the inserts based on the segmentation. Results Phase binning accuracy showed no variation for respiratory gated dataset. The calculated volumes of the phantom inserts on the phases of the 4D image dataset were typically comparable with those on the 3D static dataset as was the CT numbers. The magnitude of geometric distortion was found to be small. Conclusion 4DCT scans of a dynamic thorax phantom and patients were successfully acquired at Waikato. References 1. Starkschall G, Desai N, Balter P, et al. Quantitative assessment of Four-Dimensional Computed Tomography image acquisition quality. J. Appl. Clin. Med. Phys., Vol. 8, No. 3, 2007. 2. Jiang S B, Wolfgang J, Mageras G Sal. Quality Assurance challenges for motion-adaptive radiation therapy: Gating, Breath Holding, and Four-Dimensional Computed Tomography. Int. J. Radiation Oncology Biol. Phys., Vol. 71, No. 1, Supplement, pp. S103–S107, 2008.

P042 CT dose audit 2009–2014: Pre- and post ARPANSA NDRL

References 1. Raupach, R, Shukla, H., Amies, C., Loeffler, W., (2013) MARIS – Metal Artifact Reduction in Image Space – Technical Principles, Siemens Healthcare whitepaper 2. Markwell, T, 2015, Megavoltage Cone Beam CT with a standard Medical Linear Accelerator, PhD Thesis, QUT, Brisbane 3. Reft, C., Alecu, R., Das, I. J., Gerbi, B.J., Keall, P., Lief, E., Mijnheer, B.j., Papanikolaou, N., Sibata, C., Van Dyk, J., (2003) Dosimetric considerations for patients with HIP prostheses undergoing pelvic irradiation. Report of the AAPM Radiation Therapy Committee Task Group 63

P041 Initial assessment of 4DCT at Waikato Hospital P. Moleme, J. Egan1, K. Mugabe1 1

Waikato Regional Cancer Centre, Hamilton, New Zealand ([email protected]),

K. S. Mong1, P. U1, M. K. Badawy1 1

Medical Physics Department, Austin Health, Australia ([email protected]), ([email protected]), ([email protected]) Introduction Since the inception of NDRL in Australia in 2011, CT dose optimisation continues to be a key focus in patient dose management. At Austin Health, we decided to perform a CT dose audit to review the CT dose trend and the effect NDRL has had at this practice. Method Retrospective data collection of CTDIvol and DLP values of procedures for 2 years prior to NDRL and 3 years post NDRL were taken. The following non-contrast procedures were audited, CT brain, CT chest, CT abdomen/pelvis and CT chest/abdomen/pelvis scans. Data was collected across 4 CT scanners and their median of the spread per year was taken as the facility reference level (FRL) as defined in the ARPANSA National Diagnostic Reference Level Service Fact Sheet. The FRL was then compared with the NDRL values.

123

Australas Phys Eng Sci Med Results Preliminary results indicate a dose reduction over time since 2009 for most procedures, which may be attributed to dose optimisation, CT technology and the establishment of NDRL as a benchmark for CT dose comparison. Conclusion Regular CT dose audits are useful in managing dose trends and dose optimisation. It can be seen that since the implementation of the NDRL in 2011, there has been a positive trend towards lowering radiation dose associated with CT at this facility. References 1. ARPANSA, National Diagnostic Reference Level Fact Sheet (2013)

P043 The top 10 questions two physics PhD students asked themselves when searching for dosimetric predictors of trial treatment toxicities C. R. Moulton1, N. Yahya1 1 School of Physics, University of Western Australia, Crawley, Western Australia ([email protected]), ([email protected])

Introduction In the era of Google-orchestrated World Brain (Wells, 1938), no fleeting thoughts and questions need to be left unturned… they need to be documented! The objective of the collaboration was to distil our PhD projects down to 10 common questions when searching for dosimetric predictors of trial treatment toxicities. Method We are two PhD students who are three years through our candidatures. Our projects are on: (i) Modelling urinary toxicities in patients treated with prostate external beam radiotherapy and (ii) The occurrence of rectal bleeding in patients receiving external beam radiotherapy followed by a high-dose-rate brachytherapy boost. The challenges are from analysing trial data covering more than 900 patients across 23 institutions. Separately, we made a list of 15 questions we were most concerned about during our projects. The lists were collated and the 10 most common questions were selected. These questions were presented to ten observers who provided comments. Results The questions covered the treatment planning stage (e.g. how well the treatment planning dose reflects the dose delivered?) through to the analysis stage (e.g. how well would this model work for other trial data?). Some questions were clinically driven such as what is a consistent and justified way of assigning a patient as having a toxicity event? Conclusion Upon reflection of our experiences we were awakened by an avalanche of questions and many possible solutions. We anticipate revealing these would inspire debate, connect paradigms and advance ideas to form a consensus. References 1. Wells, H. World Brain. London. Methuen & Co. 1938.

300.00

200.00

Normal

100.00

Moderate Broad

0.00 4 Channel

20 Channel

64 Channel

Fig. 1 Comparison of SNR between coils and varying pre scan normalisation options 1

Liverpool Cancer Therapy Centre, Liverpool, Australia ([email protected]). 2Department of Medical Physics, Liverpool Hospital, Australia ([email protected]), ([email protected]). 3Ingham Institute for Applied Medical Research, Liverpool, Australia ([email protected])

Introduction Magnetic resonance imaging (MRI) for radiation therapy planning (RTP) requires good signal uniformity for precise target delineation. Sensitivity profiles of RF coils; used as many separate elements to improve SNR or as phased arrays to improve scan coverage, create significant variations in signal intensity that has to be corrected prior to image utilisation. The purpose of this study is to compare three possible RF coil set-ups used for MRI for the purpose of head and neck (H&N) planning in terms of the uniformity correction and other image quality metrics. Methods All images were acquired using a wide bore 3T Siemens Skyra MRI. A phantom (melon) was scanned utilising a T2 turbo spin echo (TSE) sequence with TE/TR = 75/15490 ms, 450 Hz/Px bandwidth, a 22 cm field-of-view, a 2/0 mm slice thickness/gap and 0.5 9 0.5 mm voxel size. Three RF coil arrangements were evaluated: two lateral 4-channel flex coils, a 20-channel and a 64-channel head coil. The effects of three pre-scan normalisation corrections (normal, moderate and broad range) were evaluated in each case. Measurements signal-to-noise ratio (SNR), contrast-to-noise-ratio (CNR) and uniformity were obtained across all slices for each set-up and correction setting. Results There was only minimal variation in image uniformity across the 3 RF coil arrangements although the best performer was the 64 channel using normal and broad range correction (92.5 %).The 64 channel coil demonstrated superior SNR and there was a dependence on correction observed in each coil (Fig. 1). The 20 channel head coil with normal correction produced the highest CNR measurements. Conclusion Intensity correction is an important part of MRI-based RT planning but its use needs careful consideration with respect to both RF coil set-up and trade-offs in other aspects of image quality.

P045 A feasibility study of using ArcCheck for patient specific QA in Stereotactic Ablative Radiotherapy (SABR) Prabhakar Ramachandran1, Adbulrahman Tajaldeen2, David Taylor1, Derrick Wanigaratne1, Karl Roozen1, Tomas Kron1

P044 Comparison of MRI image quality for radiation therapy planning of head and neck cancer considering RF coil arrangements Robba Rai1, Aitang Xing2, Lois Holloway2,3, Gary Liney2,3

123

1 Peter MacCallum Cancer Centre, Vic, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]). 2RMIT University, Vic, Australia ([email protected])

Australas Phys Eng Sci Med Introduction Stereotactic Ablative Radiotherapy (SABR) is one of the most preferred treatment techniques for early stage lung cancer. This technique has been extended to other treatment sites like Spine, Liver, Scapula, Sternum etc., This has resulted in increased physics QA time on machine. In this study, we’ve tested the feasibility of using ArcCheck as an alternative method to replace film dosimetry. Methods Twelve patients with varied diagnosis of Lung, Liver, scapula, sternum and Spine undergoing SABR were selected for this study. Pre-treatment QA was performed for all the patients which include ionization chamber and film dosimetry. The required gamma criteria for each SABR plan to pass QA and proceed to treatment is 95 % (3 %, 1 mm). In addition to this routine process, the treatment plans were exported on to an ArcCheck phantom. The planned and measured dose from the ArcCheck device were compared using four different gamma criteria: 2 %, 2 mm, 3 %, 2 mm, 3 %, 1 mm and 3 %, 3 mm. In addition to this, we’ve also introduced errors to gantry, collimator and couch angle to assess sensitivity of the ArcCheck with potential delivery errors. Results The ArcCheck mean passing rates for all twelve cases were 76.1 ± 9.7 % for gamma criteria 3 %, 1 mm, 89.5 ± 5.3 % for 2 %, 2 mm, 92.6 % ± 4.2 % for 3 %, 2 mm, and 97.6 % ± 2.4 % for 3 %, 3 mm gamma criteria. When SABR spine cases are excluded, we observe ArcCheck passing rates higher than 95 % for all the studied cases with 3 %, 3 mm, and ArcCheck results in acceptable agreement with the film gamma results. Conclusion Our ArcCheck results at 3 %, 3 mm were found to correlate well with our non-SABR spine routine patient specific QA results (3 %, 1 mm). We observed significant reduction in QA time on using ArcCheck for SABR QA. This study shows that ArcCheck could replace film dosimetry for all sites except SABR spine.

Results The sites with the highest LAR estimates were the ipsilateral and contralateral lungs, and contralateral breast for all treatment techniques. For right sided target volumes the liver also resulted in high LAR estimates, with all techniques having a LAR greater than 20 per 10,000 person-years (PY), except for mammosite with a mean LAR estimate of 13.2 per 10,000 PY. For left sided target volumes the stomach also resulted in high LAR estimates, with both whole breast and APBI having a LAR greater than 20 per 10,000 PY, and mammosite the lowest with a LAR of 8.3 per 10,000 PY. Conclusion As expected from reported clinical studies, the lungs and contralateral breast showed high LAR estimates. Results show that mammosite techniques results in the lowest risk estimate for SPC formation. References 1. L.G. Marcu, A. Santos, E. Bezak, ‘‘Risk of second primary cancer after breast cancer treatment,’’ European Journal of Cancer Care 23, 51–64 (2014). 2. U. Schneider, M. Sumila, J. Robotka, ‘‘Site-specific dose–response relationships for cancer induction from the combined Japanese A-bomb and Hodgkin cohorts for doses relevant to radiotherapy,’’ Theoretical Biology and Medical Modelling 8, 27 (2011).

P047 Improvements to the IMRT QA process using the IBA MatrixX Philip Satory1, Shahin Fattahi2 1

P046 Estimating the secondary primary cancer incidence after breast radiotherapy A. M. C. Santos1, E. Bezak2, L. G. Marcu3 1 School of Physical Sciences, University of Adelaide, Adelaide, Australia, and Department of Medical Physics, Royal Adelaide Hospital, Adelaide, Australia ([email protected]). 2 School of Physical Sciences, University of Adelaide, Adelaide, Australia. 3School of Physical Sciences, University of Adelaide, Adelaide, Australia and Faculty of Science, University of Oradea, Oradea, Romania

Introduction Induction of second primary cancers (SPCs) after breast radiotherapy (RT) has been known for some time1. Organs such as the lungs and the oesophagus have been identified as common sites for SPC formation after breast RT. The current study investigated the risk of secondary carcinogenesis associated with particular radiotherapy techniques for breast cancer; these included 3DCRT whole breast, segmented breast, partial breast and mammosite brachytherapy. Methods Seven breast cancer patients had all major organs contoured by a radiation oncologist on their planning CT images. Whole breast, segmented breast, accelerated partial breast irradiation (APBI) and mammosite boost treatment plans were generated for each patient using Pinnacle3 v 9.2 treatment planning system. Differential dose volume histograms (DVHs) were generated for critical structures: bladder, brain and CNS, breast, colon, liver, lung, mouth and pharynx, oesophagus, ovary, salivary gland, small intestine, stomach, and uterus. The lifetime attributed risk (LAR) of cancer induction was then estimated using the Schneider et. al 2 excess absolute risk models and the calculated DVHs for the above organs.

Genesis Cancer Care WA ([email protected]). 2Genesis Cancer Care WA ([email protected]) Introduction IBA’s MatrixX and OmniPro ImRT have been in use for a number of years (S. Amerio, et al. 2004). There are 3 areas that we noticed that we could improve on the vendor instructions, volume averaging, interpolation, and resolution alignment of measured and TPS data. The scope of the project was to institute improvements to the IMRT process within OmniPro ImRT software. Method (A) Rather than using 1 mm for resolution for the dose planes we used the measurement spacing (7.619 mm) divided by and integer (8) to give a spacing of 0.952 mm (limited to 3 DP by TPS). Then the difference between the closest point in the doseplanes and each of the chambers was compared. (B) The vender’s instructions say to use 7.619 mm square volume averaging. Due to institutional inertia a square of 3.8095 mm was used for volume averaging. The areas of these were compared to the area of the chamber with a diameter of 4.5 mm. (C) While it has been shown that arras have the information to completely describe the distribution (Bj¨orn Poppe et al.), this does not mean that linear interpolation is justified. Dose planes of decreasing resolutions were compared to each other to quantify the interpolations errors. Results (A) For 1 mm spacing an average linear agreement for 0.45 mm was found, for 0.952 mm spacing the linear agreement was 0.026 mm. (B) Using 7.619 mm give and area of 58 mm2, 3.8095 give and area of 14.5 mm2 and the chamber has an area of 15.9 mm.

123

Australas Phys Eng Sci Med (C)

0.952 to 1.904 give 5 % of the pixels having a greater than 1 % error and 0.33 % grater the 3 % error. This increases to 9.5 [ 1 % and 7 [ 3 % for 3.8 mm to 1.904

Conclusion With little cost these changes are worth doing.

the data using least squares polynomial regression was then used to derive the detector specific correction factors as a function of fieldsize. Results By collating published kX output correction factors for each detector investigated, a detector-specific mean over all the values obtained by a change in accelerator design, model, filtration, beam energy, beam quality, collimation (jaws, MLCs, cones), SSD, depth, and derivation was realised. Conclusion The process of averaging published output correction factors takes into account a number of aspects used to explain data discrepancies [1]. It could be particularly useful if accurate data specific to the department’s linac model, collimation, and geometric setup does not exist. References 1. Benmakhalouf, H., J. Sempau, and P. Andreo (2014). ‘‘Output correction factors for nine small field detectors in 6 MV radiation therapy photon beams: A PENELOPE Monte Carlo study.’’ Med. Phys., 41(4):041711.

Fig. 1 The lifetime attributed risk estimates for a left-sided breast targets and b right-sided breast targets References 1. S. Amerio, et al., ‘‘Dosimetric characterization of a large-area pixel-segmented ionization chamber,’’ Med. Phys. 31: 414–420 (2004). 2. OmniPro I’mRT system version 1.7b users guide (2011) 3. Bj¨orn Poppe et al., ‘‘Spatial resolution of 2D ionization chamber arrays for IMRT dose verification: single-detector size and sampling step width’’ Phys. Med. Biol. 52 (2007) 2921–2935

P048 The use of average output correction factors as a pragmatic approach for small field relative dosimetry C. Stanton1, R. Artschan1 1

Department of Radiation Oncology, Calvary Mater Newcastle, Australia ([email protected]), ([email protected]) Introduction Output factor measurements are required during beam commissioning to model the increase in dose-to-water per MU with field size due to increased collimator and phantom scatter. However, small field (\3 9 3 cm2) measurements become complicated by loss of lateral charged particle equilibrium, occlusion of direct photon beam source, and changes to beam spectra. Indeed, depending on the choice of detector, there is potential for large errors when effects such as volume averaging, energy dependence and differences in material composition are not taken into account. And whilst numerous publications provide detector-specific correction (kX) factors, values can vary depending on the method of calculation, with no consensus currently available on the values to apply under different linac, collimation, and geometric setups. As such, this work investigates the feasibility of applying mean values from presently published data, dependent on detector type and field size. Method A literature review of published kX small field output correction factors was performed for each of the solid-state detectors owned by the department, namely the SunNuclear EDGE, PTW microDiamond, and IBA EFD. Values were noted and normalised to a 3 9 3 cm2 cross-calibration field size. A best fit to the mean value of

123

P049 Atlas-Based auto-segmentation for head and neck – the road to clinical implementation K. Summerhayes1, M. Najim1, E. Flower1,2, R. Stensmyr1, J. R. Sykes3 1

Crown Princess Mary Cancer Centre, Westmead, Australia ([email protected]), ([email protected]), ([email protected]). 2Institute of Medical Physics, University of Sydney, Australia ([email protected]). 3 Blacktown Cancer and Haematology Centre, Blacktown, Australia ([email protected]) Introduction Atlas-based Auto-Segmentation has the potential to save time on the laborious task of contouring and reduce inter-observer variability [1, 2]. This study focuses on three variables available to the user in the implementation of an atlas based autosegmentation tool: (1) the optimal number of atlas subjects, (2) the finalisation method and (3) the application of post-processing workflows. Method 20 patients were manually delineated by a single Radiation Oncologist for this study. Atlas Based Segmentation was performed with three sets of atlases with 10, 15 and 20 patients respectively using a Leave-One-Out-Cross-Validation Study. This was repeated using a single-atlas match and with multiple-atlas matches (4 and 5), implementing either the Majority Vote or STAPLE label fusion strategy. Additionally, post possessing tools to smooth, remove islands and fill holes were applied to the automatic segments and the processed segments were quantitatively assessed. The Dice Similarity Co-efficient (DSC) was used to compare the performance of auto segmented structures with those contoured manually. Results Increasing the number of patients in the atlas set did not improve the performance of atlas-based auto-segmentation, with the original 10 patients producing the highest mean DSC values. The single-atlas match, with a mean DSC of 0.642, was outperformed by multiple-atlas matches, which had mean DSC of 0.704 and 0.700 for STAPLE and Majority Vote respectively. The implementation of a post processing workflow improved the mean DSC of the best performing auto-segmentations from 0.704 to 0.707 (p = 0.016, paired t-test) and qualitatively improved the contours; however, also resulted in the loss of smaller structures e.g. lenses.

Australas Phys Eng Sci Med Conclusion Best results for atlas-based segmentation were achieved with the smallest set of atlases (10 patients) with multiple-atlas matching and post processing tools implemented. Further optimisation of the atlas set is required before clinical implementation. References 1. Han, X., et al. (2008) 2. Stapleford, L.J., et al. (2010)

P050 Clinical introduction of TruView radiochromic gel S. Sylvander1, A. G. Livingstone1, C. M. Lancaster1, S. B. Crowe1 1 Cancer Care Services, RBWH, QLD, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected])

Introduction TruView radiochromic gels are commercially available (Modus Medical Devices Inc) ferrous xylenol-orange gel dosimeters (Bero et al. 2001) optimised for optical CT scanners. This study details the clinical introduction of gel dosimetry using TruView gels at the Royal Brisbane and Women’s Hospital. Method The linearity of dose response was evaluated using irradiations up to 20 Gy. The spatial stability of the dose distribution was evaluated over time. The suitability of the gel for patient-specific quality assurance was tested using a volumetric-modulated arc therapy (VMAT) treatment plan. The potential use for multiple irradiations, with previous irradiations removed as background signal in a pre-scan, was also investigated. Scanning was done at wavelengths of 633 nm and 590 nm in a Vista 15 cone beam optical scanner. Readings were acquired and processed using Vista software suite. Results Difficulties were encountered in the refrigeration of the gel, with a number of samples freezing and becoming unusable. The dose response was linear up to 20 Gy. Diffusion times were greater than advertised by vendor. The removal of existing signal by pre-scanning was feasible. Reasonable agreement was found between the gel measurement and treatment planning system dose calculation. Conclusion The dose response of the gel met clinical standards. The gel was usable to 20 Gy. Care with the storage temperature of the TruView gel needs to be observed. The shelf life of a month may present difficulties for clinical adoption. References 1. Bero, M et al. (2001) Radiochromic gel dosemeter for three-dimensional dosimetry. Radiat Phys Chem 61: 433–435

P051 Comparison of image quality for radiochromic and radiographic film S. Sylvander1, S. B. Crowe1, K. Francis1 1

Cancer Care Services, RBWH, QLD, Australia ([email protected]), ([email protected]), ([email protected]) Introduction The transition from radiographic to radiochromic film comes with a range of benefits including self-development, extended sensitive dose range, and insensitivity to visible light. Qualitative

differences in image quality can be seen in clinical use. This study investigated image quality metrics for Gafchromic EBT2 and EBT3, and Kodak EDR2 and XV2 film. Method Two experiments were designed to calculate modulation transfer function (MTF), signal-to-noise ratio (SNR), and contrast-tonoise ratio (CNR). These values were calculated from measurements made downstream of the PIPSpro QC-3 test phantom, designed for testing image quality from electronic portal imaging devices. The values were also determined using a slant edge target test on measurements made downstream of water-equivalent phantom material and a lead block. Films were scanned with an Epson V10000 using a resolution of 72 DPI. Analysis was performed using red channel data in ImageJ. The number of MU delivered to each film was selected for optimal dose response: 1000, 500 and 250 MU for EBT2 and EBT3; 313 MU for EDR2 and 65 MU for XV2. Results The results suggest that there is no significant difference in MTF between Gafchromic and radiographic film. Radiographic film did, however, exhibit superior CNR and SNR when compared to Gafchromic film. This suggests that radiographic film is better able to detect small or low contrast effects. Conclusion Clinics transitioning from radiographic to radiochromic film need to be aware that effects previously detected may be missed.

P053 Influence of ion channels on thalamocortical neuron excitation during electric stimulation – A computational model Siva Mahesh Tangutooru1, Faris Tarlochan1 1 Department of Mechanical and Industrial Engineering, Qatar University, Qatar ([email protected]), ([email protected])

Introduction Thalamocortical (TC) neurons have the relay type of neurons which respond to their inputs and then relay information to cortex. These relay neurons in their soma and dendritic membranes display various ionic channels with dynamic conductances. Simulations were performed primarily focusing on the characterization of roles of each type of channels during the action potential generation after excitation. These simulations unveiled the dynamics of firing of TC neurons, and described the interplay between the channel conductances. Method An electrophysiological model of TC neuron has been designed and simulated using the finite element method in the COMSOL Multiphysics modelling environment. A stimulation electrode was also modelled to provide necessary current stimulus needed for the neuron excitation. TC neuron and electrode models were coupled by positioning in a homogeneous isotropic volume conductor mimicking the tissue medium. The current balanced equation (Equation) by a Hodgkin-Huxley type of model was used to model the neuron’s plasma membrane [1]. The channels that were incorporated with the membrane are summarized in the Table. C

dV ¼ ðIsti  IT  IH  INa  IKdr  IKs  INal  IKl Þ dt

Results The electrophysiological properties of TC relay neurons after electrical stimulation were investigated by characterizing the role of each individual channel’s current. The calcium current IT generated low threshold Ca2+ spike (Figure). The transient K+ current IKdr delayed the rate of rise and reduced the peak amplitude of the Ca2+ spike, but IKs contributed majorly to the repolarization

123

Australas Phys Eng Sci Med of the Ca2+ spike. Addition of IH generated an apparent afterhyperpolarization after the Ca2+ spike and thus created a potential for rhythmic oscillations of Ca2+ spikes. These oscillations also depend on the status of leakage conductances that effects input resistance of the cell and apparently the membrane potential.

P054 Imaging under uncertainty: double-slit experiment performed with digital X-ray sensor used in clinical radiology service C. E. C. Teixeira1,2,3, E. M. Melo2, K. K. Santos2, D. Rezende2,4, A. R. Rodrigues3 1

Laborato´rio Multidisciplinar, Centro Universita´rio do Estado do Para´, Brazil. 2Centro de Cieˆncias Biolo´gicas e da Sau´de, Universidade da Amazoˆnia, Brazil ([email protected]), ([email protected]). 3Nu´cleo de Medicina Tropical, Universidade Federal do Para´, Brazil ([email protected]), ([email protected]). 4DIMAGEM Diagno´sticos por Imagem, Brazil ([email protected])

A single calcium spike generated followed by the electrical stimulation of 1 volt Ionic channels and their associated maximum values of TC neuron membrane Abbr. Channel currents

Maximum values

IT

T-type calcium current

0.0001 cm/s (permeability)

IH

Hyperpolarized activated cation current

0.0005 S/cm2 (conductance)

INa IKdr

Sodium current Delayed rectifier potassium current

0.03 S/cm2 (conductance) 0.003 S/cm2 (conductance)

IKs

Slow potassium current

0.0007 S/cm2 (conductance)

INal

Sodium leakage current

0.0000095 S/cm2 (conductance)

IKl

Potassium leakage current

0.00005 S/cm2 (conductance)

C

Membrane capacitance

1 lF/cm2 (capacitance)

V

Membrane potential

65 mV (resting membrane potential)

Introduction The physical properties of radiological signals influence contrast, resolution and noise, i.e. digital image quality. One important physical property of radiological signals is the wave-particle duality, evidenced at diverse levels of matter (see References). In this work, we performed the double-slit experiment in a radiology service room using digital radiography equipment to evidence wave-particle duality influencing X-rays detection by digital sensors in conditions mimicking routine in clinical radiology. Method We used four lead plates (24 cm width, 15 cm height and 4 mm thickness), being one plate without any slit, one plate with a central slit, and two plates with two slits separated by 5 and 15 mm, respectively. Each slit had 1 mm width, 5 cm height and 4 mm thickness. The plates were placed 1000 mm from the emission source and 100 mm from the digital sensor, mimicking patients’ positioning during radiographic exam. X-rays were produced using 51 kV, 500 mA, 20 mAs. Exposition time was 0.04 s. Results The frequency of X-rays detection at digital sensor space is shown in the Figure below (A: no slit; B: one slit; C and D: two slits separated by 5 and 15 mm, respectively). The interference pattern was observed when slits were separated by 5 mm (Figure C). Conclusion Particle-wave duality might influence the quality of radiographic digital images.

Conclusion This study determined the electrophysiological properties and the mutual dependence of current channels in TC relay neurons. References McIntyre, Cameron C., et al. ‘‘Cellular effects of deep brain stimulation: model-based analysis of activation and inhibition.’’ Journal of neurophysiology 91.4 (2004): 1457–1469. Acknowledgment This research was made possible by the NPRP grant #5-457-2-181 from the Qatar National Research Fund (a member of Qatar Foundation). The statements made herein are solely the responsibility of the authors.

123

References 1. Young, T. (1802) The Bakerian lecture: On the theory of light and colours. Phyl Trans R Soc Lond, 92: 12–48.

Australas Phys Eng Sci Med 2. Zeilinger, A. et al. (1988) Single- and double-slit diffraction of neutrons. Rev Mod Phys, 60: 1067–1073. 3. Tonomura, A. et al. (1989) Demonstration of single-electron buildup of an interference pattern. Am J Phys, 57(2): 117–120. 4. Carnal, O & Mlynek, J (1991) Young‘s double-slit experiment with atoms: a simple atom inteferometer. Phys Rev Lett, 66: 2689–2692. 5. Nairz, O et al. (2002) Experimental verification of the Heisenberg uncertainty principle for fullerene molecules. Phys Rev A, 65: 032109.

P055 Modeling and verification of four dosimetrically matched Elekta linear accelerators for 6MV VMAT treatments D. P. Truant1, S. Arumugam1, B. Beeksma1, J. Begg1, J. Hellyer1, R. Short, P. Vial1, T. Young, G. Goozee 1

Cancer Therapy Centre, Liverpool and Campbelltown Hospital, Australia ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected]), ([email protected])

Introduction Volumetric Modulated Arc Therapy (VMAT) is the next generation of Intensity Modulated Radiation Therapy (IMRT) techniques in radiotherapy [1,2,3]. Using a single treatment planning system (TPS) beam model for multiple linear accelerators across two departments greatly simplifies work flow but requires precise beam matching and accurate dosimetric verification, particularly if complex treatments such as Head-and-Neck VMAT are to be safely implemented [4,5]. Method Following beam matching across four Elekta linacs (two Synergy and two Versa HD linacs, all with Agility multileaf collimators) a single 6MV photon beam was modelled in the Pinnacle3 TPS. Seven Head-and-Neck VMAT treatments were generated and delivered on all machines. Dosimetric accuracy was verified using ion chamber measurements, and gamma analysis with the ArcCHECK diode array dosimeter, EPID dosimetry and Gafchromic film (qualitative only). Results Agreement in 6MV photon beams across the four machines and TPS were within 1 % for Percent Depth Dose curves, 2 % for flatness and symmetry and 0.5 % for field output factors. Accurate MLC rounded leaf end modelling was verified using abutting MLC fields [6]. For the Head-and-Neck VMAT clinical treatments, across all machines the average in-phantom ion chamber local dose difference was -0.6 % (from measured to TPS) with standard deviation (SD) 2.6 %. Across all machines ArcCHECK array average global gamma pass rates for 3 %3 mm/2 %2 mm criteria was 98.1 %/ 89.8 % (SD 1.9 %/5.8 %). EPID fluence average global gamma pass rates for 3 %3 mm/2 %2 mm criteria was 97.4 %/89.6 % (SD 4.5 %/ 7.1 %). Conclusion Successful beam matching of 6MV photons with a single TPS model across four machines was achieved within international guideline tolerances [7]. The dosimetric accuracy results for clinical H&N VMAT treatments were also within international guidelines [5].

A single TPS model was validated and clinically implemented across multiple Elekta machines for complex VMAT techniques. References 1. Otto, K. (2008), Volumetric modulated arc therapy: IMRT in a single gantry arc. Med. Phys. 35, 1, 310–317. 2. Bedford, J. L. (2009), Treatment planning for volumetric modulated arc therapy. Med. Phys. 36, 5128. 3. Bortfeld, T. Webb, S. (2008) Single-Arc IMRT? Phys. In Med. Bio. 54, 1, N9–N20. 4. Verbakel, W. et al. (2009), Volumetric Intensity Modulated Arc Therapy vs. Conventional IMRT in Head-and-Neck Cancer: A Comparative Planning and Dosimetric Study. Int. Journ. of Rad. Onc. Bio. Phys. 74, 1, 252–259 5. Code of Practice for the Quality Assurance and Control of Volumetric Modulated Arc Therapy (2015) Netherlands Commission on Radiation Dosimetry. 6. Rice, J. R. (2014) Optimisation of the rounded leaf offset table in modelling the multileaf collimator leaf edge in a commercial treatment planning system. Journ. Of App. Clin. Med. Phys. 15, 6, 128–137 7. Technical Report Series 430: Commissioning and Quality Assurance of Computerised Planning Systems for Radiation Treatment of Cancer (2004) International Atomic Energy Agency.

P056 Developing real-space filtering techniques for image enhancement in computed tomography reconstruction Michael Williams1, Stephen Bosi1,2, Konstantin Pavlov1,3 1 School of Science and Technology, University of New England, Australia ([email protected]). 2Institute of Medical Physics, University of Sydney. 3School of Physics and Astronomy, Monash University, VIC, Australia

Introduction Cone Beam CT using 2D panel detectors is finding increasing application in medical imaging, image guidance etc. However, 2D detectors are more susceptible to noise from scattered radiation than detectors used in more traditional fan-beam CT (Pan et al. 2008). Feldkamp filtered back-projection is one of the most efficient and most commonly-used reconstruction algorithms (Wang et al. 2008, Hsieh et al. 2013). Part of the algorithm involves adjusting amplitudes of different image spatial frequency components by applying a suitable filter in Fourier space (Fourier filter or ‘‘kernel’’). Typically, clinical CT systems deal with image noise by modifying the Fourier filter response to adjust the attenuation or enhancement of different frequency components in the image. This choice is a compromise between minimising image noise and maximising contrast of tissue boundaries, because attenuating high frequencies reduces the contrast of both noise and sharp boundaries. Method The conventional approach has focused on improving Fourier filter functions. Our research is exploring methods of selectively pre-enhancing tissue boundaries in CT scan projections in realspace before Fourier filters are applied. The algorithms exploit prior knowledge of properties of tissue features and boundaries to enhance boundary contrast versus noise. Subsequent application of a noise reducing Fourier filter would attenuate noise but pre-enhanced boundaries and features would be partially protected from excess attenuation, improving signal-to-noise and contrast-to-noise ratios.

123

Australas Phys Eng Sci Med Results Experiments indicate the Sobel filter applied to projections in real space is promising. Conclusion Experiments are ongoing and results will be presented at the time of conference. References 1. Hsieh, J., B. Nett et al. (2013). Current Radiology Reports 1(1): 39–51. 2. Pan, X., J. Siewerdsen, P. J. et al. (2008). Medical Physics 35(8): 3728–3739. 3. Wang, G., H. Yu and B. De Man (2008). Medical Physics 35(3): 1051–1064.

compared to the TPS-predicted dose for a number of IMRT test cases using a volumetric DVH analysis. Conclusion The feasibility of this system has been demonstrated for IMRT test deliveries. EPID-based 2D pre-treatment QA results can be translated into 3D patient dose for clinically relevant patient specific IMRT QA. References 1. King et al. (2012). Development and testing of an improved dosimetry system using a backscatter shielded electronic portal imaging device. Medical Physics 39(5), 2839–2847.

P057 Converting measured 2D dose discrepancies into 3D patient dose using pre-treatment EPID images for IMRT QA

Development of a clinical model for predicting urethral stricture following external beam radiotherapy of the prostate: urethral dose-length as an important predictor

R. David1,2, B. Zwan1,2, J. Hindmarsh1, E. Seymour1, K. Sloan1, C. Lee1,2, P. Greer2,3

N. Yahya1, A. Ebert1,2, A. Steigler3, A. Kennedy2, D. J. Joseph2,3, J. W. Denham3

Central Coast Cancer Centre, CCLHD, Gosford, NSW. 2School of Mathematical and Physical Sciences, University of Newcastle, NSW, Australia ([email protected]), ([email protected]). 3Calvary Mater Hospital, Newcastle, NSW, Australia

1

1

Introduction The purpose of pre-treatment quality assurance (QA) is to ensure the correct dose is delivered to the patient. This typically involves comparison between a measured and predicted dose distribution in a simple phantom. Using this methodology it is difficult to assess the delivery in terms of parameters which are clinically relevant to the patient. There are two main reasons for this: (1) measurements are performed in an abstract phantom which is not representative of the patient and (2) measured dose cannot be assessed volumetrically in terms of the delineated target volume and organs at risk. In this work we propose a system which translates 2D dose measurements into a 3D dose distribution in the patient’s planning CT using pre-treatment EPID images. Method EPID images were acquired in-air on a Varian Clinac for IMRT deliveries. For each IMRT beam the acquired images were converted to dose planes at 10 cm depth in a water phantom and compared to the dose planes predicted by the Eclipse TPS (King et al. 2012). A 2D ratio map constructed for each beam was used to modify the 3D beam dose in the planning CT, taking into account the divergence of the beam. The modified 3D dose for each beam was then summed to give the total modified 3D patient dose. This modified dose could then be imported back into the TPS and compared to the planned dose distribution using a dose-volume-histogram (DVH) analysis.

Results 3D patient dose distributions have been reconstructed and

123

School of Physics, University of Western Australia, Western Australia, Australia ([email protected]). 2 Department of Radiation Oncology, Sir Charles Gairdner Hospital, Western Australia, Australia. 3School of Medicine and Public Health, University of Newcastle, New South Wales, Australia. 3School of Surgery, University of Western Australia, Western Australia, Australia Introduction Urethral stricture is commonly associated with brachytherapy but a rare toxicity in external beam radiotherapy of the prostate. With the introduction of dose-escalation in the treatment of prostate, the urethra is similarly exposed to higher doses, thus increasing the risk. A predictive model for stricture incidence would help guide use of dose-escalation. In this study, clinical and dosimetric factors, including the urethral dose-length, were studied for their relationship to urethral stricture and a predictive model was developed. Method Following the TROG03.04-RADAR trial, data for 754 participants treated with external beam radiotherapy to the prostate were available for analysis. The urethral dose-length, describing the length of the urethra receiving dose above a threshold, were derived from the 3-dimensional dose matrix of each treatment plan. The predictive capability of the urethral dose-length, clinical factors and medication intake was assessed univariately. Multivariate logistic regression was used to derive a clinical prediction model; the model was internally validated. Results 13 patients had stricture. The length of urethra receiving C72 Gy (L72) was the strongest dosimetric predictor in univariate analysis. The final prediction model for urethral stricture included pre-treatment bowel condition (odds ratio (OR) = 3.74, 95 % confidence interval (CI) = 1.04–13.45, p = 0.044) and L72 (OR = 1.56/ cm, CI = 1.13–2.14/cm, p = 0.006). The optimism-corrected area under the receiver operating curve of the model was 0.73. Conclusion The urethral length receiving high dose is an important predictor of urethral stricture. This result was obtained using a very crude surrogate for the urethra and delivered urethral dose. External validation is required.