Room: Exhibit Hall | Forum 6
Purpose: Organ motion might become a potential source of uncertainty in MR-driven radiotherapy when the total examination time in the MR-scanner is extensive or during time consuming replanning in the MR-linac. The aim of this study was to investigate the dosimetric impact of anatomical changes from organ motion for prostate cancer patients using MRI.
Methods: Two large field of view T2-weighted sequences were acquired at the start (MR1) and end (MR2) of the protocol for ten prostate cancer patients. The sequences were separated by approximately 30 minutes, equivalent to the time for re-planning at the MR-linac. MR1 was used for organ delineation and treatment planning was performed using a synthetic CT (sCT, MriPlannerTM, Spectronic Medical) generated from MR1. The treatment plan was copied to a second sCT generated from MR2 and the dose was recalculated, resulting in two dose-distributions D1 and D2. MR2 was deformable registered towards MR1 using Elastix in MICE Toolkit (NONPI Medical). Using this registration, D2 was warped to the geometry of MR1 and compared against D1. The inverse deformation was applied to MR1 delineations for investigation of anatomical changes.
Results: Organ position and/or volume change was observed for all patients. Bladder had a mean volume change of 46cc (27cc-68cc) and rectum volume changed between -5cc to 13cc. Nine patients had under-dosage of the PTV Dmean (-0.5Gy to 0.0Gy). Four patients showed under-dosage of more than 2 Gy to the PTV D98%, (max: -6Gy). Large variation was seen for the rectum D15% (-15Gy to 6Gy).
Conclusion: In MR-driven radiotherapy, when data from different sequences acquired over time are used in combination, or when a â€œplan of the dayâ€? MR-linac approach is used, anatomical changes can be of major dosimetric impact. Anatomical changes can force the prostate out of the PTV boundaries, increasing the risk of target under-dosage.
Funding Support, Disclosures, and Conflict of Interest: Funded by VINNOVA, The Swedish Innovation Agency.