Room: AAPM ePoster Library
Purpose: MR residual distortions, although small, may have a great impact on treatment plans with a sharp dose gradient, such as spine radiosurgery (SRS) plans. To treat SRS patients with a 1.5T MR-Linac, it is necessary to assess the impact of MR distortions and develop a solution to mitigate this impact.
Methods: Nine CT-based SRS treatment plans (prescribed to 18 or 24 Gy) with assumed underlying MR distortion were collected for this study. A previously validated polynomial model was applied to measured MR-Linac geometric distortion data, creating distortion vectors at voxel level. The distortion vectors were applied to deform the GTV and cord/cauda contours to generate the presumably actual locations of the GTV and cord/cauda. At the same time, we expanded the original GTV and cord/cauda contours using a variable margin expansion algorithm, which was based on the union of the original and deformed contours through differential mathematical morphology. A new plan was reoptimized by using the expanded contours, while keeping planning parameters and other contours the same as the original plans.
Results: Impact of MR distortions was case specific. Of the 9 plans, one plan had underdose to the target due to the distortion (GTV mean dose decreased from 25.1Gy to 23.9Gy); and four plans had increased cord/cauda maximum dose (6.2%, 8.6%, 12.2% and 22.7%), of which one increased from 11.6 Gy to 14.2 Gy, resulting in significant overdose to the spinal cord. After reoptimization using the expanded contours, that GTV mean dose increased to 24.2Gy while keeping the same cord dose level; and the maximum cord dose of those 4 plans fell back towards the original clinical goals.
Conclusion: We demonstrated a variable margin expansion approach to account for MR image distortion in SRS planning. The approach should be implemented if SRS patients will be treated with MR-Linac.
Funding Support, Disclosures, and Conflict of Interest: This work was funded by Elekta.