Room: Karl Dean Ballroom C
Purpose: MRI-only treatment planning using synthetic CT (sCT) is under rapid development however quality assurance methods are required to validate synthetic CT accuracy for clinical workflows. This work aims to develop and investigate a verification method for sCT scans using comparison of modelled and measured fraction 1 transit EPID images.
Methods: For a multi-center prospective clinical trial for prostate patients undergoing treatment with MRI-only planning, sCT scans were generated with a hybrid method combining atlas based and voxel based approaches. Seven-field clinical IMRT plans were generated using MRI contours and sCT dose calculations. Predicted absolute grayscale transit EPID images were generated using a comprehensive physics-based analytical model. Measured transit EPID images were acquired during treatment at fraction 1. 2D gamma analysis and percentage mean absolute signal difference (S) were used for comparison. Comparisons were made between: 1) EPID image predictions through sCT scans and corresponding CT scans for 17 patients; 2) EPID image predictions through sCT scans and measured transit EPID images from fraction 1 for 6 patients.
Results: For the comparison of modelled transit images (n=17) the average (Â±1 SD) of the gamma pass rates was 86.5% (Â±14.1%), 94.1% (Â±9.7%) and 97.6% (Â±5.2%) with 2%,2mm, 3%,3mm, and 4%,4mm criteria respectively. The average (Â±1 SD) of S was 1.4% (Â±1.2%). For the comparison of modelled and measured transit images (n=6) the average (Â±1SD) of the gamma pass rates was 75.7% (Â±6.2%), 91.8% (Â±4.8%) and 96.6% (Â±3.8%). The average (Â±1SD) of S was 2.1% (Â±1.6%). Some individual field cases were found to have large differences due to path-length differences between plan and treatment however average patient results had higher pass-rates.
Conclusion: Verification of accuracy of sCT scans is feasible using transit EPID image prediction model compared to measured transit EPID image at the first fraction.
Not Applicable / None Entered.