J Gastrointest Surg (2010) 14:1969–1973 DOI 10.1007/s11605-010-1283-y

ORIGINAL ARTICLE

Development of a Multimodal Tumor Model for Porcine Liver Anna Rethy & Thomas Langø & Jenny Aasland & Ronald Mårvik

Received: 18 May 2010 / Accepted: 28 June 2010 / Published online: 24 July 2010 # 2010 The Society for Surgery of the Alimentary Tract

Abstract In our efforts to develop a guidance system for laparoscopic liver surgery, we are working towards a live animal tumor model. The objective of this study was to establish the tumor model for live porcine liver, visible on both computed tomography (CT) and ultrasound images. The tumor model was created by injecting a mixture of agarose, sephadex, and glycerol. Together with water, the mixture was heated to bring its components into solution. Once heating was complete, methylthionine chloride and CT contrast were added. Using laparoscopic ultrasound guidance, the tumor model mixture was injected into in vivo porcine liver. The resulting model tumors were radiolucent, visible on both CT and conventional X-ray. They appeared as hyperechoic lesions on ultrasound images. Compared to the CT images, the model tumors in the ultrasound images showed good correspondence in size. We conclude that our tumor model, due to its clearly identifiable nature on multiple imaging modalities, is a valuable tool for further studies on laparoscopic ultrasound (2D and 3D) and navigated ultrasound in laparoscopic surgery of the liver and other organs in a pre-clinical set-up. Keywords Tumor model . Laparoscopy . Ultrasound . Navigation . Three-dimensional ultrasound

Introduction

A. Rethy : J. Aasland : R. Mårvik Future Operating Room Project, St. Olavs Hospital, Trondheim, Norway

On a mission to improve the safety and efficacy of laparoscopic liver surgery, navigation systems are being developed to combine and integrate real-time laparoscopic ultrasound (LUS) images with preoperative computed tomography (CT) or magnetic resonance images.1–5 To develop and evaluate the usability, possible clinical benefits, and the accuracy of these systems, a life-like tumor model in a live animal set-up is both beneficial and necessary. The purpose of this study was to develop a tumor model in an in vivo porcine liver tissue that is visible on multiple imaging modalities. The tumor model would have the following characteristics:

T. Langø (*) Department of Medical Technology, SINTEF, 7465 Trondheim, Norway e-mail: [email protected]

& & &

Grant support See Acknowledgements section. A. Rethy Medical Faculty, Norwegian University of Science and Technology (NTNU), Trondheim, Norway

J. Aasland : R. Mårvik Surgical Department, St. Olavs Hospital, Trondheim, Norway R. Mårvik National Center for Advanced Laparoscopic Surgery, St. Olavs Hospital, Trondheim, Norway

&

Discrete, well circumscribed lesions Lesions visible on both ultrasound and CT images Lesion consistency similar to consistency of liver parenchyma Easily identifiable on gross examination

We aimed to achieve a tumor consistency similar to that of liver parenchyma in order to facilitate unbiased targeting of the tumors with tracked surgical instruments in further studies, i.e., the surgeon should be able to rely exclusively on the navigation system for sense of direction and location

1970

of tumor in guidance of procedures and should not be influenced by sensing a different consistency with the surgical instrument. Furthermore, to facilitate further experiments using the tumor model in studies on LUS-guided liver resection, we have compared sample images from CT and ultrasound to demonstrate the feasibility of the model.

J Gastrointest Surg (2010) 14:1969–1973

then drawn up into a 10-ml syringe. Two milliliters of CT contrast (Omnipaque 270 mg/ml) was drawn up into another syringe. The contents of the syringes were mixed, with the help of a three-way catheter to prevent air from entering the mixture, yielding our tumor model mixture. Several syringes with the tumor model mixture were prepared and then maintained at a temperature of 65°C in a hot water bath.

Materials and Methods Ex Vivo Model The animal experiments in this study were approved by the national committee for research on animals. In addition, the study protocol was approved by our hospital research scientific board. Initial Model As a first attempt, the tumor model presented by Restrepo et al.6 was used as inspiration; KY jelly (water-soluble lubricant, Johnsen & Johnsen) with different amounts of CT contrast fluid was used to create tumor mimicking lesions in the liver. 1. Components mixed in open container; resulting mixture was full of air bubbles (a) KY jelly+50% CT contrast (Omnipaque 270 mg/ml, GE Healthcare) 2. Components mixed using two syringes connected by a two-way catheter to prevent air bubble formation (a) KY jelly+50% CT contrast (Omnipaque 270 mg/ml) (b) KY jelly+30% CT contrast (Omnipaque 270 mg/ml) (c) KY jelly+20% CT contrast (Omnipaque 270 mg/ml)

An ex vivo model was created to establish solidification time of tumor model mixture. A bovine liver was cut into nine 10×10-cm pieces. The liver pieces were placed individually into plastic bags and maintained at 38°C in a water bath. The tumor model mixture was maintained in a water bath at 65°C and then injected approximately 2 cm deep into liver pieces. Syringes were left in place to prevent leakage. Fig. 1 shows a photo of the transected tumor model and the liver parenchyma-like characteristics. In Vivo Model With the pig under general anesthesia, three laparoscopic ports were made: one supraumbilical and two in the left upper quadrant. The sites were chosen in order to have proper access to the liver. The needle was placed through the skin and guided into the liver under laparoscopic visualization (Fig. 2a–c). Using ultrasound guidance, the needle tip was advanced to a depth of 1.5 to 2.0 cm in the liver (Fig. 2b–c). Care was taken to avoid injuring vascular or biliary structures with the needle. A 2–3-ml bolus of the tumor model mixture was rapidly injected through an 18-

Final Model Our final tumor model, based on Scott et al.,7 was created by injecting a mixture of 6 g of agarose, 6 g of sephadex, and 14 ml of glycerol with enough tap water to make a 200-ml volume. The mixture was then heated to 95°C using a microwave oven to bring its components into solution. Microwave energy on a high-power setting was applied for a total of approximately 3 min, for 30 s at a time to prevent mixture from overflowing. Once heating was complete, 2 ml of methylthionine chloride (10 mg/ml) was added. The mixture was divided equally and transferred to four glass jars and sealed with airtight lids. Gradual cooling at room temperature resulted in a solid material with a tough gelatin-like consistency. The jars were stored for up to 4 weeks at room temperature prior to use. When needed for an experiment, the jars were individually reheated in a microwave oven for 30 s at a time until the gelatin was completely re-liquefied. Eight milliliters of the mixture was

Fig. 1 Photo of a transected model tumor from an ex vivo bovine liver.

J Gastrointest Surg (2010) 14:1969–1973

1971

Acquisition of Images Computed Tomography The operating room (OR) where the experiments were performed contains a roof mounted C-arm fluoroscopy unit (Axiom Artis dTa system, Siemens, Germany; Fig. 3a). In addition to regular fluoroscopy and X-ray imaging, it has cone beam CT (CBCT) functionality. By acquiring approximately 400 X-ray images while rotating 220° around the OR table, it is able to reconstruct CT-like images. The images are reconstructed on a workstation (Leonardo, Siemens, Germany) and can be visualized in the OR. This CBCT imaging modality has some quality limitations compared to regular multislice CT,8 but the intraoperative imaging possibilities provide great experimental possibilities (and patient logistics). Ultrasound The ultrasound images were acquired using an LUS probe (OL531, Hitachi, Japan; Fig. 3b). Results In Vivo—KY Jelly Model Mixture was injected under LUS guidance using an 18-gauge epidural needle. Tacosil® was used for hemostasis and to

Fig. 2 Laparoscopic ultrasound (LUS)-guided injection of the liver model tumors. a Overview showing two needles used to inject tumors and one in progress. Laparoscope and LUS probe are used for guidance. b Laparoscope video image showing the LUS probe tip and one needle. c LUS image of hyperechoic tumor with color flow imaging. The needle is causing the shadow in the image (in part of the tumor). Color flow is used to avoid blood vessels.

gauge epidural needle into the hepatic parenchyma. The needle was left in place for 45 min to prevent extravasation of the tumor model mixture. The puncture site was cauterized to achieve adequate hemostasis.

Fig. 3 a The C-arm used to acquire 3D cone beam computed tomography data in the experiments. b Laparoscopic ultrasound probe used in the experiments to guide the placement of the tumors and to view the final tumor models.

1972

J Gastrointest Surg (2010) 14:1969–1973

Fig. 4 Sample images of in vivo model tumors. a, b Computed tomography slices showing two model tumors. c, d The corresponding tumors seen in laparoscopic ultrasound images.

Fig. 5 Volume rendering (thresholded) of a cone beam computed tomography volume showing five tumors in the liver. a Approximately anteroposterior view. b Volume slightly tilted for better view of the tumors.

J Gastrointest Surg (2010) 14:1969–1973

1973

prevent leakage of the injected material. The resulting tumors had poorly defined borders, poor visibility on both modalities, and leakage of the injected material could not be prevented sufficiently. The presence or absence of air did not have a significant impact on the resulting tumor models.

provided by the video laparoscope, provides precision and thus added safety to minimally invasive surgery. Our tumor model due to its multimodal visibility can be used to further develop and perfect navigation systems.

Ex Vivo—Agarose Model

Conclusion

The livers were cut open one at a time at 10-min intervals. Optimal solidification time of tumor model mixture was found in the fourth liver; therefore, optimal solidification time was established at 40 min.

Our tumor model, being equally well visible on both ultrasound and CT, creates a set-up for developing guidance systems in controlled animal trials in order to improve their accuracy and feasibility. We believe that the model can be a valuable tool for further studies on navigation systems and LUS, both 2D and 3D, in laparoscopic surgery of the liver and other organs in a pre-clinical set-up.

In Vivo—Agarose Model Resulting tumors were well circumscribed with clearly defined borders. They were clearly visible on both CT (Fig. 4a–b) and ultrasound (Fig. 4c–d). Consistency was constant, permanent, and similar to the consistency of the liver parenchyma. With the 3D CT capability of the C-arm in our OR set-up, we can acquire volume renderings such as the one shown in Fig. 5 for better overview of all tumors injected. This feature will be exploited in further studies.

Acknowledgements This study was supported by SINTEF (Trondheim, Norway), The Ministry of Health and Social Affairs of Norway, through the National Center for 3D Ultrasound in Surgery (Trondheim, Norway), project 196726/V50 eMIT (Enhanced minimally invasive therapy, FRIMED program), and the Future Operating Room project at St. Olavs Hospital (Trondheim, Norway). We would like to thank Kirsten Rønning and Anne Karin Wik for valuable help during the experiments in the OR.

References Discussion The tumor model was developed primarily with navigation in mind. The intraoperative CBCT was used out of convenience of availability and also as a gold standard for comparison to ultrasound images (particularly in further studies). Navigation systems have the potential to improve the safety and efficacy of laparoscopic surgery. Ultrasound integrated with preoperative CT can help the understanding of the LUS images in correspondence with surrounding anatomy. Solberg et al.5 have shown that image fusion techniques make it easier to perceive the integration of two or more volumes in the same display (monitor) than mentally fusing the same volumes presented in their own separate displays. The ultrasound data will show updated information that the surgeon relies on during surgery, while the advantages from CT, such as better overview and understanding of the anatomy and pathology, are displayed simultaneously. Herein lies the importance of the tumor model being visible on multiple imaging modalities. Surgical instruments (with integrated tracking technology) can be visualized in these volumes. This opens up the possibility for the laparoscopic surgeon to visualize the exact location of the surgical instruments in relation to the preoperative CT images combined with real-time LUS images. Having an additional image, to the standard image

1. Baumhauer M, Feuerstein M, Meinzer HP, Rassweiler J. Navigation in Endoscopic Soft Tissue Surgery: Perspectives and Limitations. Journal of Endourology. 2008;22(4):751–66. 2. Estépar RSJ, Stylopoulos N, Ellis RE, Samset E, Westin CF, Thompson C, Vosburgh K. Towards scarless surgery: An endoscopic ultrasound navigation system for transgastric access procedures. Computer Aided Surgery 2007;12(6):311–24. 3. Feuerstein M, Reichl T, Vogel J, Schneider A, Feussner H, Navab N. Magneto-optic tracking of a flexible laparoscopic ultrasound transducer for laparoscope augmentation. LNCS, Proc Int Conf Med Image Comput Computer-Assisted Intervention (MICCAI) 2007;4791(Part I):458–66. 4. Hildebrand P, Schlichting S, Martens V, Besirevic A, Kleemann M, Roblick U, Mirow L, Burk C, Schweikard A, Bruch HP. Prototype of an intraoperative navigation and documentation system for laparoscopic radiofrequency ablation: first experiences. Eur J Surg Oncol 2008;34(4):418–21. 5. Solberg OV, Langø T, Tangen GA, Mårvik R, Ystgaard B, Rethy A, Hernes TAN. Navigated ultrasound in laparoscopic surgery. Minim Invasive Ther Allied Technol (MITAT) 2009;18(1):36–53. 6. Restrepo JI, Stocchi L, Nelson H, Young-Fadok TM, Larson DR, Ilstrup DM. Laparoscopic ultrasonography: a training model. Dis Colon Rectum 2001;44:632–7. 7. Scott DM, Young WN, Watumull LM, Lindberg G, Fleming JB, Rege RV, Brown RJ, Jones DB. Development of an in vivo tumormimic model for learning radiofrequency ablation. J Gastrointest Surg 2000;4:620–5. 8. Eide KR, Ødegård A, Myhre HO, Lydersen S, Hatlinghus S, Haraldseth O. DynaCT during EVAR—A Comparison with Multidetector CT. Eur J Vasc Endovasc Surg 2009;37:23–30.