Pediatr Radiol DOI 10.1007/s00247-016-3604-0

ORIGINAL ARTICLE

How accurate is size-specific dose estimate in pediatric body CT examinations? Boaz Karmazyn 1,2 & Huisi Ai 3 & Paul Klahr 4 & Fangqian Ouyang 5 & S. Gregory Jennings 2

Received: 24 October 2015 / Revised: 1 February 2016 / Accepted: 1 March 2016 # Springer-Verlag Berlin Heidelberg 2016

Abstract Background Size-specific dose estimate is gaining increased acceptance as the preferred index of CT dose in children. However it was developed based on non-clinical data. Objective To compare the accuracy of size-specific dose estimate (SSDE) based on geometric and body weight measures in pediatric chest and abdomen CT scans, versus the more accurate SSDE (mean SSDE based on water-equivalent diameter). Materials and methods We retrospectively identified 50 consecutive children (age <18 years) who underwent chest CT examination and 50 children who underwent abdomen CT. We measured anteroposterior diameter (DAP) and lateral diameter (DLAT) at the central slice (of scan length) of each patient and calculated DAP+ LAT (anteroposterior diameter plus lateral diameter) and DED (effective diameter) for each patient. We calculated

* Boaz Karmazyn [email protected]

1

Department of Radiology, Riley Hospital for Children, 705 Riley Drive, Room 1053, Indianapolis, IN 46202, USA

2

Department of Radiology and Imaging Sciences, Indiana University School of Medicine, Indianapolis, IN, USA

3

Department of Radiation Oncology, Indiana University School of Medicine, Indianapolis, IN, USA

4

CT Clinical Science, Philips Healthcare, Highland Heights, OH, USA

5

Department of Biostatistics, Indiana University School of Medicine, Indianapolis, IN, USA

the following in each child: (1) SSDEs based on DAP, DLAT, DAP+LAT, DED, and body weight, and (2) SSDE based on software calculation of mean water-equivalent diameter (SSDE adopted standard within our study). We used intraclass correlation coefficient (ICC) and Bland– Altman analysis to compare agreement between the SSDEs and SSDE. Results Gender and age distribution were similar between chest and abdomen CT groups; mean body weight was 37 kg for both groups, with ranges of 6–130 kg (chest) and 8–107 kg (abdomen). SSDEs had very strong agreement (ICC>0.9) with SSDE. SSDEs based on DLAT had 95% limits of agreement of up to 43% with SSDE. SSDEs based on other parameters (body weight, DAP, DAP+LAT, DED) had 95% limits of agreement of up to 25%. Conclusion Differences between SSDEs calculated using various indications of patient size (geometric indices and patient weight) and the more accurate SSDE calculated using proprietary software were generally small, with the possible exception for lateral diameter, and provide acceptable dose estimates for body CT in children. Keywords Abdomen . Chest . Children . Computed tomography dose index volume . Size-specific dose estimate

Introduction Children are more sensitive than adults to the potential adverse effects from ionizing radiation, with higher radiation dose to smaller body size using the same technique [1]. Since the introduction of the ALARA principle (dose as low as reasonably achievable), pediatric

Pediatr Radiol

radiologists have investigated different methods of estimating pediatric patient doses from CT scans [2, 3]. Computed tomography dose index volume (CTDIvol) is the most frequently used parameter for this estimation because it is reported by any state-of-the-art CT scanner [4]. However CTDIvol is problematic because it measures not the radiation dose to the patient, but the radiation dose delivered to a specific standardized phantom. At a given CTDIvol, a baby might receive more than twice the radiation dose of an adult [5, 6]. This problem prompted the development of a new measure, called size-specific dose estimate (SSDE). The SSDE was introduced in report 204 of the American Association of Physicists in Medicine (AAPM) [5]. The SSDE is calculated as a product of CTDIvol and a sizespecific conversion factor. SSDE is the mean specific radiation dose to the patient calculated for one central slice location of the scan length. SSDE has been evaluated for adult and pediatric CT scans of the chest, abdomen and pelvis and has gained acceptance as a measure of CT scan dose [7–11]. One of the potential limitations of SSDE is that the size-specific conversion factor is based on the geometry of the patient torso and not on X-ray attenuation. AAPM report 220 introduced the SSDE as a more robust method of measuring dose. This method is based on water-equivalent diameter, which includes consideration of both the physical size of the patient and the degree of X-ray attenuation by the patient. The SSDE is calculated as a mean of the product of water-equivalent diameter and size-specific conversion factor in each axial CT slice [12]. SSDE calculation, however, is more difficult than using body weight or purely geometric measurements and requires proprietary software. The purpose of our study is to retrospectively compare, in children who underwent routine chest and abdomen CT scans, how SSDE calculated using various geometric size metrics and body weight varies from the more accurate SSDE.

Materials and methods Patients This study complies with the Health Insurance Portability and Accountability Act and was approved by our institutional review board with a waiver of informed consent. We retrospectively identified 50 consecutive pediatric patients who underwent chest CT and 50

others who underwent abdomen and pelvis CT during the period of April 2014 to August 2014 at our tertiary care children’s hospital. We did not include repeat studies on the same child or CT studies that included both chest and abdomen in a single scan. From medical records we noted age, gender and body weight. From the CT dose report we documented the CT scanner estimated dose index volume based on a 32-cm phantom (CTDIvol). CT scan protocol All scans were performed using iCT 128 (Philips Medical Systems, Cleveland, OH) in helical mode with 4-cm collimation using a slice thickness of 4 mm, increment of 3 mm, a reconstruction filter (kernel) B, and a matrix of 512×512. Rotation time was 0.33–0.5 s. Pitch range was 0.9–1.1. We used automated current selection, which was determined by the calculated water-equivalent diameter from the scout view (Table 1). Both craniocaudal and angular tube current modulations were performed during all scans. Geometric-based diameters The child’s anteroposterior diameters (DAP) and lateral diameters (DLAT) at the mid-reconstructed CT slice along the scan length of the child were measured in the workstation using digital calipers (Fig. 1). The effective diameter (DED) was pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi calculated as follows: DAP
DLAT . We used a program (Cede, Philips Healthcare, Best, the Netherlands) to calculate in each child examination the mean water-equivalent diameter (Fig. 1). Mean water-equivalent diameter is the mean of the waterequivalent diameters calculated for each slice based on the calculated mean Hounsfield units [12]. Calculation of size-specific dose estimate The SSDE was calculated as a product of scanner’s estimated CTDIvol and a conversion factor corresponding to the tables provided by the AAPM report 204 [5]. For calculation of the SSDE based on body weight, we used the table from the study of Khawaja et al. [13]. The SSDE was calculated based on the AAPM report 220. We used the Cede program as follows: SSDE ¼  n ∑z¼1 CTDIvol ðzÞ
f Dw ðzÞÞ =n, where fDw is the size-specific conversion factor dependent on slicespecific water-equivalent diameter.

Pediatr Radiol Table 1 Body weight

CT scan parameters kVp

a

Reference parameters Phantom diameter

Chest mAs

Abdomen mAs

<20 kg

80

16 cm

60

90

20–40 kg

100

20 cm

40

60

>40 kg

100

32 cm

120

180

kVp kilovoltage peak, mAs milliamperes a

Automated current selection program adjusted the mAs based on the estimated water-equivalent diameter (calculated from the scout film attenuation) and the reference phantom diameter

of agreement was defined as follows: 0–0.2=poor agreement, 0.3–0.4=fair agreement, 0.5–0.6=moderate agreement, 0.7–0.8=strong agreement and >0.8=very strong agreement. We used Bland–Altman analysis to compare agreement between each SSDE and SSDE. The mean of the SSDE measurements and SSDE were calculated and then plotted against the percentage difference of the two estimates. Horizontal lines were drawn at the mean percentage difference and at the 95% limits of agreement, which are defined as the mean percentage difference +/- 1.96 times the standard deviation (SD) of the percentage differences.

Statistical analysis

Results

Descriptive statistics included mean and range for data with a normal distribution or as median and interquartile range (IQR) for non-normally distributed data. As a reference standard for body X-ray attenuation and CT dose, we used mean water-equivalent diameter and SSDE, respectively. Agreement between D ED to mean water-equivalent diameter and between each SSDE and SSDE were calculated as intraclass correlation coefficients (ICC). For each of the body diameter indices and SSDEs, we evaluated the median and interquartile range (IQR) of the deviation from mean waterequivalent diameter and SSDE, respectively. Deviation was calculated as (X–Y)/Y*100%, where X is denoted as each body diameter’s indices, or SSDEs, Y is denoted as mean water-equivalent diameter or SSDE. Degree

Demographics

D ED had very strong agreement with mean waterequivalent diameter for the chest CT scans (ICC=0.91) as well as for the abdomen CT scans (ICC=0.97). In

Fig. 1 Measurements of the geometric and water-equivalent diameters in a post-contrast abdomen CT scan in a 9-year-old boy. a Anteroposterior and lateral diameters are measured in the mid-scan axial slice. b Water-

equivalent diameter is calculated at each axial CT scan slice based on mean attenuation, as seen in the graphic representation that overlays the multiplanar mid-coronal scan

For chest CT, we included 21 girls and 29 boys, with mean age of 9.3 years (range 6 months to 18 years), and with mean body weight 37.1 kg (range 5.8 kg to 129.5 kg). For abdomen CT, we included 19 girls and 31 boys, with a mean age of 9.7 years (range 6 months to 17.7 years) and a mean body weight of 37.7 kg (range 7.5 kg to 106.6 kg). Effective and water-equivalent diameters

0.95 0.99 0.94

DAP anteroposterior diameter, DAP+LAT anteroposterior + lateral diameters, DED effective diameter, DLAT lateral diameter

0.99 0.93 0.92 0.93 0.96 0.96 Intraclass correlation coefficient

0.98

3.7% (–0.3%, 8.6%) −2.2% (–4.3%, 0.6%) 5.6% (3.1%, 9.4%) −0.1% (–3.7%, 1.8%) 6.2% (1.4%, 10.2%) −0.5% (–3.7%, 4.0%) 6.0% (0.1%, 11.6%) −1.5% (–4.8%, 4.2%) −1.6% (–6.4%, 4.7%) Median deviation (interquartile range)

0.0% (–4.0%, 4.2%)

4.6 (3.8, 5.4) 7.5 (6.7, 8.0) 4.7 (3.9, 5.4) 7.3 (6.6, 7.9) 4.8 (3.9, 5.4) 7.5 (6.6, 8.3) 4.8 (4.0, 5.6) 6.9 (6.2, 7.6) 4.6 (4.1, 5.3)

Abdomen Chest Chest Abdomen Chest

Abdomen

DLAT DAP Body weight

The concept of SSDE is gaining increased acceptance as a CT scan patient dose estimate [5, 7, 8, 11, 14, 15]. Compared to the CTDIvol, SSDE takes into account the size of the child by the use of a size-specific conversion factor. This conversion factor was derived from experimental and Monte Carlo data normalized to patient water-equivalent diameter [5]. In our study, we evaluated the accuracy of SSDE calculated using body weight, and geometrical diameters in children who underwent routine chest and abdomen CT scans. We used the methods described in AAPM Task Group report 220 to estimate the water-equivalent diameter taking X-ray attenuation within the scanned area into account. We adopted the SSDE (mean of all SSDEs calculated at each slice location of the entire CT scan) as our standard because it is the most accurate estimate of the actual SSDE available. The SSDE has a few potential deficiencies. One is that the estimation of the body size is obtained at midscan and may not represent accurately the average size of the entire body. Another is the use of a geometric estimate of body size, the effective diameter, as a surrogate for a patient’s X-ray attenuation [8, 12]. X-ray attenuation is the fundamental physical process affecting the absorption of X-rays and is thus more relevant than geometric patient size in determining the radiation dose

Characteristics of each of the size-specific dose estimates (SSDE) in the chest (C) and abdomen (A) CT scans

Discussion

Table 2

Bland–Altman findings suggest good agreement between the SSDE values and SSDE, with low mean percentage differences (0–9%) and narrow 95% limits of agreement (less than +/- 25%, with the exception of SSDELAT [SSDE based on DLAT] chest) (Fig. 2). In the chest, SSDELAT had 95% limits of agreement up to 43% with one outlier of 134% (Table 3).

D AP + LAT

Bland–Altman

Chest

Abdomen

The median of SSDE was 4.8 mGy (IQR 3.9 to 5.4 mGy) in the chest and 6.8 mGy (IQR 6.5 to 7.5 mGy) in the abdomen. Each of the SSDEs had very strong agreement (ICC>0.9, P<0.05) with SSDE for both chest and abdomen CT scans (Table 2).

Chest

DED

Size-specific dose estimate

Median SSDE (interquartile range)

Abdomen

comparison with mean water-equivalent diameter, DED deviated on average by 7.4% (range –3.9% to 20.5%) in the chest CT scans and –3.4% (range –16.7% to 4.4%) in the abdomen.

7.3 (6.6, 7.9)

Pediatr Radiol

Pediatr Radiol

Fig. 2 Bland–Altman analysis compares agreement between average sizespecific dose estimates (SSDEs) and SSDE base on software calculation of mean water-equivalent diameter in the chest (a–e) and abdomen (f–j). The solid line indicates the mean percentage difference and the dashed lines, the 95% limits of agreement. a, f SSDEBW – SSDE based on body weight; b, g

SSDEAP – SSDE based on anteroposterior diameter; c, h SSDELAT– SSDE based on lateral diameter; d, I SSDE AP+LAT – SSDE based on anteroposterior diameter plus lateral diameter; e, j SSDEED – SSDE based on effective diameter. In (c), one outlier for SSDELAT of 134% was beyond the range of the scale and was not included

Pediatr Radiol Table 3 The mean differences and limits of agreement of each size-specific dose estimate (SSDE) measurement and SSDE in chest and abdomen CT scans

SSDE

Chest

Abdomen

Mean % difference

95% limits of agreement

Mean % difference

95% limits of agreement

Body weight

2%

−16%/13%

0%

−16%/16%

DAP

1% 2% 0% 2%

−11%/12% −39%/43% −11%/11% −12%/9%

7% 7% 7% 5%

−11%/25% −11%/24% −9%/22% −11%/21%

DLAT DAP+LAT DED

DAP anteroposterior diameter, DAP+LAT anteroposterior + lateral diameters, DED effective diameter, DLAT lateral diameter

to the child. For example, in the chest, the lungs do not attenuate the X-ray beam as much as solid organs, and therefore the effective diameter in the chest might overestimate attenuation [8, 12]. A few studies have demonstrated that automated calculation of SSDE is a reliable method [8, 9, 12, 13]. Wang et al. [16] compared geometric and attenuation-based diameters to mean water-equivalent diameter and found that DED based on different phantom types and sizes representing abdominal sizes from newborn to adult deviated on average by –1.5% (range –2% to –0.7%). In this study, on different phantoms representing the adult chest, DED deviated from mean water-equivalent diameter on average by 13.4% (range 7.5% to 21.5%) [16]. Our study, based on chest and abdomen CT scans, demonstrated comparable results to the Wang publication, with an average deviation of 7.4% (range –3.9% to 20.5%) in the chest and –3.4% (range –16.7% to 4.4%) in the abdomen CT scans. The DED overestimation of chest attenuation and underestimation of abdomen attenuation is expected because the air in the lungs has less attenuation than water, and the solid organs in the abdomen attenuate X-rays more than water. The Wang et al. [16] study cited by the AAPM report 220 compared only body diameters but not SSDE. We are not aware of any other study that compared SSDE based on different geometric diameters and weight to the more robust CT dose measure, SSDE. In a study by Leng et al. [8], an automated calculation of the SSDE in adult chest, abdomen and pelvis CT scans was compared to automated calculation of mid-scan SSDE based on water-equivalent diameter (SSDE). They found that the maximum deviation of mid-scan SSDE was up to 6% compared with SSDE. Our study, using the more robust SSDE estimation based on the entire CT scan data, also showed very strong agreement (ICC≥0.95) between most SSDE measurements and SSDE. With the exception of SSDELAT, the mean percentage difference was close to 0% and the 95% limits of agreement for each SSDE to SSDE for both the chest and abdomen CT scans were within ±25%. SSDELAT had 95% limits of agreement of up to 43% and

one outlier with the percentage difference of 134%. We could not find an explanation for the outliers. Our overall results show good agreement with the AAPM report 204, which suggested SSDE accuracy within 20% [2]. SSDEBW had 95% limits of agreement of within ±16% for the chest and the abdomen water-equivalent diameter substantiate its use as a simplified method to calculate SSDE [3]. The advantage of using body weight is that there is no need to select a mid-scan location and perform measurements in one or two diameters with calculations, all of these steps take time and introduce interobserver variability [13]. This study has several limitations related to the retrospective nature of the study, a small sample size, and the use of automated calculations of mean water-equivalent diameter, which might vary among vendors. In addition, mean waterequivalent diameter calculation might be less accurate in large body sizes when the entire skin-to-skin image is not displayed.

Conclusion We found that except for SSDELAT in the chest, SSDEs based on mid-scan geometrical diameter have good agreement with the more accurate method, SSDE, with 95% limits of agreement of within ±25%. Compliance with ethical standards Conflicts of interest None

References 1. 2.

Slovis TL (2002) The ALARA concept in pediatric CT: myth or reality? Radiology 223:5–6 Berdon WE, Slovis TL (2002) Where we are since ALARA and the series of articles on CT dose in children and risk of long-term cancers: what has changed? Pediatr Radiol 32:699

Pediatr Radiol 3. 4.

5.

6. 7. 8.

9.

10.

Brody AS, Frush DP, Huda W et al (2007) Radiation risk to children from computed tomography. Pediatrics 120:677–682 Huda W, Mettler FA (2011) Volume CT dose index and dose-length product displayed during CT: what good are they? Radiology 258: 236–242 Boone JM, Strauss KJ, Cody DD et al (2015) Size-specific dose estimates (SSDE) in pediatric and adult body CT examinations. AAPM report No. 204. https://www.aapm.org/pubs/reports/RPT_ 204.pdf. Accessed 15 Feb 2016 McCollough CH, Leng S, Yu L et al (2011) CT dose index and patient dose: they are not the same thing. Radiology 259:311–316 Strauss KJ, Goske MJ (2011) Estimated pediatric radiation dose during CT. Pediatr Radiol 41:472–482 Leng S, Shiung M, Duan X et al (2015) Size-specific dose estimates for chest, abdominal, and pelvic CT: effect of intrapatient variability in water-equivalent diameter. Radiology 276:184–190 Christianson O, Li X, Frush D et al (2012) Automated size-specific CT dose monitoring program: assessing variability in CT dose. Med Phys 39:7131–7139 Brady SL, Mirro AE, Moore BM et al (2015) How to appropriately calculate effective dose for CT using either size-specific dose estimates or dose-length product. AJR Am J Roentgenol 204:953–958

11.

12.

13.

14. 15.

16.

Christner JA, Braun NN, Jacobsen MC et al (2012) Size-specific dose estimates for adult patients at CT of the torso. Radiology 265: 841–847 McCollough C, Bakalyar DM, Bostani M et al (2015) Use of water equivalent diameter for calculating patient size and size-specific dose estimates (SSDE) in CT. AAPM Report No. 220. https://www.aapm.org/pubs/reports/RPT_220.pdf. Accessed 15 Feb 2016 Khawaja RD, Singh S, Vettiyil B et al (2015) Simplifying sizespecific radiation dose estimates in pediatric CT. AJR Am J Roentgenol 204:167–176 Strauss KJ (2014) Dose indices: everybody wants a number. Pediatr Radiol 44:450–459 Pourjabbar S, Singh S, Padole A et al (2014) Size-specific dose estimates: localizer or transverse abdominal computed tomography images? World J Radiol 6:210–217 Wang J, Christner JA, Duan X et al (2012) Attenuationbased estimation of patient size for the purpose of size specific dose estimation in CT. Part II. Implementation on abdomen and thorax phantoms using cross sectional CT images and scanned projection radiograph images. Med Phys 39: 6772–6678