Evaluation of the True Composite and Field-by-Field QA Methods in Detecting Systematic MLC Positioning Errors

Abstract

A modern approach to creating radiation therapy treatments involve development of a computer-based treatment plan is done with ‘inversed planning’, where a treatment planning system (TPS) generates a set of complex fluence segments, satisfying user-defined objectives. The purpose of this work is to compare the detection sensitivity to a systematic error in True Composite (TC) and field-by-field (FBF) methods of patientspecific Quality Assurance (QA) test of Volumetric Modulated Arc Therapy (VMAT) plans. The assessment was performed for complex (3 dose-level prescriptions) clinically delivered head-and-neck VMAT plans, characterized by large treatment fields and a combination of both uniform and highly nonuniform dose distribution regions.

A total of 40 beams from 10 previously treated patient plans, created using RayStation® 10A treatment planning system (TPS), were evaluated with Gamma analysis. A set of patient-specific QA plans was created for delivery with FBF method onto a three-dimensional cylindrical QA device ArcCheck (Sun Nuclear): one error-free baseline plan, and three plans with systematically introduced multi-leaf collimator (MLC) positioning errors of 1, 1.5, and 2mm offsets, generated through custom scripts. Next, the same set of QA plans was recreated for the TC delivery method. All error-free QAs were measured using ArcCheck with both FBF and TC methods. Gamma analysis was performed to compare the error-free measurements and error-induced dose plans using three criteria sets 2%/2mm, 3%/2mm, and 3%/3mm with 10% dose threshold and global normalization.

The Gamma passing rates decreased consistently with increasing MLC error shifts and with stricter criteria for both TC and FBF methods, with FBF method offering higher sensitivity to induced errors. For TG-218 recommended Gamma criteria of 3%/2mm, the average passing rates dropped to 84% or lower and to 90% for FBF and TC, respectively. The difference between the methods was not statistically significant as indicated by p-values >0.05 of the Independent Sample t-test.

While both TC and FBF techniques were able to detect the induced MLC positioning errors, FBF demonstrated consistently higher sensitivity through lower Gamma passing rates in individual QA tests. The FBF technique may also be more useful in pinpointing the specific source of errors.