Purpose: EPID-based patient treatment â€œin-vivoâ€? verification is conventionally performed post-treatment delivery. Modern stereotactic treatments now deliver very high doses very rapidly to the patient rendering this approach ineffective. This study develops and demonstrates feasibility of a novel system (WatchDog 2.0) to verify patient treatment delivery in real-time at every control point for improving patient safety in radiation treatment.
Methods: Prior to real-time verification, transit EPID images were predicted at every control point for a TrueBeam linac with aS1200 EPID (Varian Systems). Two treatment plans (a VMAT with two arcs and IMRT with five fields) were utilised to predict transit EPID images and these plans were delivered through (1) air and (2) a thoracic phantom (CIRS, USA). During real-time verification, transit EPID images (~13.5 Hz) containing encoded image properties (control point index, gantry and collimator angles, MLC and jaw positions and MUs delivered) were continuously acquired using the WatchDog 2.0 system. The transit EPID images (averaged over each control-point interval) were compared with predicted images at each control point in regard to a 3%-3mm gamma analysis (10% threshold).
Results: The mean and standard deviation of a 3%-3mm gamma analysis between predicted and corresponding acquired transit EPID images was measured at 98.7%Â±5.4 [60.3%~100.0%] for VMAT and 98.8Â±1.2 [67.5%~100.0%] for IMRT in air delivery, included 6 and 39 control points less than 90.0% due to discrepancies of imaging properties, respectively. Averages of image property discrepancies were measured at <0.1Â° (Collimator angle), <0.2Â° (Gantry Angle), <0.35 mm (MLC Leafs) and <0.2 MUs (300 MUs delivered) by comparison with log-file analysis.
Conclusion: This study successfully developed and demonstrated feasibility of a novel system for control-point based real-time treatment delivery verification. It is evident that the WatchDog 2.0 system has the potential improving patient safety in radiation therapy.