2267 - Pre-Clinical Evaluation of Stopping-Power Ratio Prediction for Proton Therapy Planning Based on Dual-Energy CT: An End-to-End Validation

J. Duan, J. Zhu, C. Tao, Q. Liu, and S. Zhang; Shandong Cancer Hospital and Institute, Shandong First Medical University and Shandong Academy of Medical Sciences, Shandong Provincial Key Medical and Health Laboratory of Pediatric Cancer Precision Radiotherapy (Shandong Cancer Hospital), Jinan, China

Purpose/Objective(s): Proton therapy planning faces a challenge in determining stopping-power ratio (SPR) from CT data due to inherent uncertainty. An innovative software from a technology company was applied to derive SPR maps directly from dual-energy CT (DECT) images. This study aims to assess the accuracy of the software platform in predicting proton ranges and perform an end-to-end validation for clinical implementation. Materials/

Methods: Model 062M Electron Density Reference Phantom (CIRS, Norfolk, VA, USA) and animal tissues, including a sheeps head, pork lean, fat, heart, liver and lung were used in this study. To maintain tissue integrity and reusability, the sheeps head and lung was fixed in two plexiglas container using 10% formalin for greater than 24 hours. Other tissues were cut in pieces respectively and combined to fill a 10 cm×10 cm ×10 cm plastic box. A single-energy CT (SECT) scan (120 kVp, 1.5 mm slice thickness, Br38 kernel) was obtained. Immediately, a DECT was acquired using energies of 80/140 kVp or 100/140 kVp (1.5 mm slice thickness, Qr40 kernel). All images were exported to syngo.via workstation for generating the SPR map. Both the SECT and SPR datasets were then imported into the treatment planning system (TPS). The theoretical SPR value and DECT-SPR value were compared for each plug in the CIRS phantom. For validation, brag peaks at 180MeV, 200MeV and 220MeV proton beams, the cumulative effect on traversing pencil beams was independently evaluated by the Giraffe MLIC measurements for each CIRS phantom plug and animal tissue. The R80 and R90 water-equivalent depths (WEDs) for each specimen were measured, and then compared to TPS WEDs from DECT-SPR. In addition, an in-house software same to DirectSPR^TM was also used to create DECT-SPR for comparison with the DECT-SPR from DirectSPR^TM. Finally, dose validations were performed using a 2D ion chamber array to con?rm the accuracy of DECT-SPR planning.
Results: The difference between the DECT-SPR values generated by DirectSPR^TM data sets and theoretical SPR values are very close (<±2%), excluding lung plug (5~6%). The TPS WEDs from DECT-SPR closely matched measured WEDs with minor discrepancies, excluding lung plug and lung tissue. The measured WEDs and TPS WEDs differences in CIRS phantom plugs is less than 2%, except lung plug (6.67%). The differences in sheeps head, pork lean, fat, heart, liver and lung are 1.54%, 1.32%, 1.09, -0.71%, -0.41%, -8.07%, respectively. The DECT-SPR values generated from our in-house software closely (<±1%) matched the DECT-SPR values generated from DirectSPR^TM excluding lung plug (approximate ±2%). For dose validations, the gamma pass rates (3%, 3 mm) are greater than 95%.
Conclusion: This study showed that SPR maps created by DirectSPR^TM from dual-energy CT can successfully be used in a treatment planning system to generate proton dose distributions and that these predictions agree well with measurements.