Potential dosimetric error in the adaptive workflow of a 1.5 T MR-Linac from patient movement relative to immobilisation systems

Abstract

AbstractIn magnetic resonance- (MR-) based adaptive workflows for an MR-linac, the treatment plan is optimized and recalculated online using the daily MR images. The Unity MR-linac is supplied with a patient positioning device (ppd) using pelvic and abdomen thermoplastic masks attached to a board with high-density components. This study highlights the dosimetric effect of using this in such workflows when there are relative patient-ppd displacements, as these are not visualized on MR imaging and the treatment planning system assumes the patient is fixed relative to the ppd. The online adapted plans of two example rectum cancer patients treated at a Unity MR-linac were perturbed by introducing relative patient-ppd displacements, and the effect was evaluated on plan dosimetry. Forty-eight perturbed clinical adapted plans were recalculated, based on online MR-based synthetic computed tomography, and compared with the original plans, using dose-volume histogram parameters and gamma analysis. The target volume covered by the prescribed dose ($${\text{D}}_{\text{p}\text{r}\text{e}}$$ D pre ) and by at least 107% of $${\text{D}}_{\text{p}\text{r}\text{e}}$$ D pre varied up to − 1.87% and + 3.67%, respectively for 0.5 cm displacements, and to − 3.18% and + 4.96% for 2 cm displacements; whilst 2%–2 mm gamma analysis showed a median value of 92.9%. The use of a patient positioning system with high-density components in a Unity MR-based online adaptive treatment workflow can introduce unrecognized errors in plan dosimetry and it is recommended not to use such a device for such treatments, without modifying the device and the workflow, followed by careful clinical evaluation, or alternatively to use other immobilization methods.