Room: AAPM ePoster Library
Purpose: MRI-based treatment planning benefits from excellent soft-tissue contrast without ionizing radiation. However, geometric distortion is of great concern. This study aims to determine the dosimetric impact of geometric distortion on photon and proton treatment plans with various dose gradients by using measurements from two commercial MRI systems.
Methods: Second-order polynomial models were created based on measured 3D geometric distortion maps from two MRI systems (dedicated MRI simulator and MR-linac). Ten patients were selected for each treatment modality: photon and proton. Disease sites include brain, pelvis, abdomen, and spine. Modeled geometric distortions were applied to the planning CT and structure sets to mimic MRI-based treatment planning excluding additional errors in converting MRI to synthetic CT. Nominal plans with intensity modulation were generated by optimizing on the distorted CT using clinical planning objectives. Absolute dose gradient was calculated for each voxel of dose map from the nominal plan. Verification plans were then generated to calculate the perturbed doses on the undistorted CT. Percentage deviation of the perturbed dose from the nominal dose was calculated and plotted as a function of distortion and percentage dose gradient.
Results: The mean geometric distortions in the field of view were 0.64 mm and 0.77 mm for two MRI systems. In the scenario of typical geometric distortion (=2mm) in clinical MRI systems, the median dose deviations for photon plans were in the range of 0.18% to 1.35%, and 0.17% to 1.1% for each system, respectively. For proton plans, the maximum median dose deviations were in the range of 0.19% to 2.02%, and 0.18% to 2.37% for each system, respectively.
Conclusion: The deviation in dose due to geometric distortion was almost doubled when comparing proton plans to photon plans. When geometric distortion is kept below 2mm, the median dose deviation is =2.5% for common proton plans.