The first experimental implementation of real-time dose-guided MLC tracking using an MRI-linac

Intrafraction motion can cause dose errors that are not corrected by geometry-based tracking. Real-time dose-guided adaptive radiation therapy (ART) has been proposed to address this limitation but has not been experimentally demonstrated. This study reports the first experimental implementation of dose-guided MLC tracking.

Dose-guided MLC tracking was integrated with an MRI-linac. MRI guidance was used to monitor a motion phantom and adaptation was performed by accumulating the desired and delivered doses based on the target motion and optimising MLC leaf positions to minimise dose errors. Tracking performance was measured for a single moving target and a multi-target setup undergoing differential motion using film and plastic scintillation dosimeters (PSD). Dose-guided MLC tracking was compared to geometry-guided MLC tracking and no tracking. End-to-end latency of the tracking system was also measured.

The average 2%/2mm gamma-pass rates for the single target setup were 90.2% for dose-guided tracking, 91.8% for geometry-guided tracking, and 42.6% for no tracking. Mean PSD dose differences from the static reference were 2.4% for dose-guided tracking, 2.0% for geometry-guided tracking, and 26.8% for no tracking. The average gamma-pass rates for the multi-target setup were 86.7% for dose-optimised tracking, 77.0% for geometry-guided multi-target tracking, and 44.5% for no tracking. The end-to-end latency of dose-guided MLC tracking was 545 ± 16 ms.

Real-time dose-guided ART was experimentally implemented for the first time. Dosimetric accuracy was comparable to geometry-guided tracking. Limitations were identified which can be addressed to fully realise the benefits of dose-guided ART on a clinical system.