Exhibit Hall | Forum 7
Purpose: Although proton therapy is prominently used for maximizing OAR sparing due to zero exit dose, inter-fraction anatomical changes could significantly degrade target coverage or increase OAR dose. Therefore, besides daily pre-treatment CBCT imaging for patient setup alignment, repeating CTs for dosimetric verification purposes (vCT) are regularly acquired. Multiple commercial CBCT-to-CT synthetic algorithms are available, but evaluations of their performances are currently limited. This study aims to assess their potentials in proton therapy quantitatively.
Methods: Ten nodal-involved prostate cases treated with intensity-modulated proton therapy were studied. The Velocity and MIM software, which respectively use deformation-based and scatter-correction-based CBCT-to-CT algorithms, were used for synthetic CT (synCT) generation. The proton plan was forwardly calculated for each case on synCTs and vCT acquired the same day. The difference of selected dose metrics for CTV, soft tissue (rectum and bladder), and bony OAR (left/right femur heads) between synCTs and vCT were evaluated and derived with p-values. The synCT of one solid water phantom case with varied geometric scenarios was visually examined for artifact and discrepancy.
Results: The average differences of CTV D95% were 0.30±0.89% for Velocity_synCT-vCT and -0.43±1.04% for MIM_synCT-vCT. The D0.1cc differences of soft tissue OAR were -1.11±6.06Gy for Velocity_synCT-vCT and -0.02±7.22Gy for MIM_synCT-vCT. The D0.1cc differences of bony OAR were -0.51±0.96Gy for Velocity_synCT-vCT and 0.76±2.15Gy for MIM_synCT-vCT. Statistical significances were found for CTV D95% (p=0.01) and bony OAR D0.1cc (p=0.01) between Velocity_synCT-vCT and MIM_synCT-vCT. Regarding phatom examination, Velocity_synCT showed fewer noticeable imaging artifacts and limitation to reproduce scenarios on CBCT compared to MIM_synCT.
Conclusion: The deformation-based synCT algorithm has smaller standard deviations for the selected dose metrics with less noticeable artifacts. In comparison, the scatter-correction-based algorithm is superior in reproducing geometric change on CBCT. For CTV and bony OAR, no difference of the evaluated dose accuracy between Velocity_synCT and MIM_synCT were revealed.