Room: AAPM ePoster Library
Purpose: MRI-guided adaptive radiation therapy (MR-guided ART) offers incredible potential for reducing sensitive normal tissue dose and increasing target coverage on a daily basis. However, accurate dose accumulation presents a significant challenge when considering the dose delivered to date. This work sought to develop an MR-guided ART dose accumulation pipeline, first benchmarked in phantom and then applied to patient cases, to support the integration of MR-guided ART in the clinic.
Methods: A novel MR-compatible pelvic end-to-end phantom (PETE) with variable air-filled rectal and fluid-filled bladder components was simulated in several conditions. Stereotactic body radiation therapy (SBRT) prostate plans were generated for each condition using a 0.35 Tesla MRI with electron density from a corresponding CT for Monte Carlo-based dose calculation. An intensity-based deformable image registration (DIR) dose accumulation pipeline was developed in MIM to accumulate doses for three different status conditions (rectal volumes of 0cc, 90cc, 150cc). As metrics for DIR performance, the mean distance to agreement (MDA) and Dice similarity coefficient (DSC) were evaluated between initial and deformed embedded phantom conditions. The pipeline was then applied to patient cases including MR-guided prostate and pancreas SBRT plans.
Results: Visual assessment of the deformations in PETE revealed good agreement. The MDA and DSC of two phantom DIRs (accumulating 3 statuses) for the planning target volume (PTV) was <0.96mm and >0.94 (bladder: <3.01mm and >0.87), respectively. Dose was accumulated in a prostate patient to a maximum dose of 43.81 Gray, with PTV MDA <1.35mm and DSC >0.86. Dose accumulation in pancreatic cancer patients presented challenges in areas of large deformations in anatomy and variations in scan extent, introducing uncertainties near the periphery.
Conclusion: An automated dose accumulation pipeline was developed to support MR-guided ART. Further validation in patient cases and optimization for complex anatomy such as pancreas is warranted.
Funding Support, Disclosures, and Conflict of Interest: The submitting institution holds research agreements with Philips Healthcare, ViewRay, Inc., and Modus Medical. Research partially supported by the National Cancer Institute of the National Institutes of Health under Award Number R01CA204189. The PI is on the Philips Healthcare Advisory Board.