Premium
Toward real‐time verification for MLC tracking treatments using time‐resolved EPID imaging
Author(s) -
Zwan Benjamin J.,
Caillet Vincent,
Booth Jeremy T.,
Colvill Emma,
Fuangrod Todsaporn,
O'Brien Ricky,
Briggs Adam,
O’Connor Daryl J.,
Keall Paul J.,
Greer Peter B.
Publication year - 2021
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1002/mp.14675
Subject(s) - multileaf collimator , imaging phantom , image guided radiation therapy , collimator , computer science , sabr volatility model , medical imaging , nuclear medicine , tracking (education) , dosimetry , radiation treatment planning , linear particle accelerator , computer vision , artificial intelligence , radiation therapy , mathematics , medicine , beam (structure) , optics , physics , radiology , volatility (finance) , psychology , pedagogy , stochastic volatility , econometrics
Purpose In multileaf collimator (MLC) tracking, the MLC positions from the original treatment plan are continuously modified to account for intrafraction tumor motion. As the treatment is adapted in real time, there is additional risk of delivery errors which cannot be detected using traditional pretreatment dose verification. The purpose of this work is to develop a system for real‐time geometric verification of MLC tracking treatments using an electronic portal imaging device (EPID). Methods MLC tracking was utilized during volumetric modulated arc therapy (VMAT). During these deliveries, treatment beam images were taken at 9.57 frames per second using an EPID and frame grabber computer. MLC positions were extracted from each image frame and used to assess delivery accuracy using three geometric measures: the location, size, and shape of the radiation field. The EPID‐measured field location was compared to the tumor motion measured by implanted electromagnetic markers. The size and shape of the beam were compared to the size and shape from the original treatment plan, respectively. This technique was validated by simulating errors in phantom test deliveries and by comparison between EPID measurements and treatment log files. The method was applied offline to images acquired during the LIGHT Stereotactic Ablative Body Radiotherapy (SABR) clinical trial, where MLC tracking was performed for 17 lung cancer patients. The EPID‐based verification results were subsequently compared to post‐treatment dose reconstruction. Results Simulated field location errors were detected during phantom validation tests with an uncertainty of 0.28 mm (parallel to MLC motion) and 0.38 mm (perpendicular), expressed as a root‐mean‐square error (RMS Error ). For simulated field size errors, the RMS Error was 0.47 cm 2 and field shape changes were detected for random errors with standard deviation ≥ 2.5 mm. For clinical lung SABR deliveries, field location errors of 1.6 mm (parallel MLC motion) and 4.9 mm (perpendicular) were measured (expressed as a full‐width‐half‐maximum). The mean and standard deviation of the errors in field size and shape were 0.0 ± 0.3 cm 2 and 0.3 ± 0.1 (expressed as a translation‐invariant normalized RMS). No correlation was observed between geometric errors during each treatment fraction and dosimetric errors in the reconstructed dose to the target volume for this cohort of patients. Conclusion A system for real‐time delivery verification has been developed for MLC tracking using time‐resolved EPID imaging. The technique has been tested offline in phantom‐based deliveries and clinical patient deliveries and was used to independently verify the geometric accuracy of the MLC during MLC tracking radiotherapy.