Investigating a correlation between MLC positional errors and IMRT QA passing rate

Abstract

Patient-specific quality assurance (QA) using measurement-based techniques has been the standard method used in many institutions as an essential step to ensure the accuracy of the actual treatment. In this work, we studied the use of the trajectory log files in specifying the influence of the linear accelerator multi-leaf collimators (MLC) positional errors in the passing rates of our measurement-based QA. In the study, we analyzed sixty-five QA treatment plan results including plans for head and neck, lungs, pelvises, craniospinal, and total body irradiation. Eclipse treatment planning system with Anisotropic Analytical Algorithm (AAA) dose calculation engine was the system used to generate all of our treatment plans. All plans were delivered on Varian® linear accelerator UNIQUE which is equipped with 120 Millennium MLC. Trajectory log files generated during treatment delivery were analyzed by a software developed using MATLAB. We compared the planned versus the actual delivered fluence from the recorded log files. Direct subtraction of both fluences allowed us to calculate the degree of match between both fluences. We called it the MLC matching percentage. Measurement-based QAs for all IMRT patients were done using Octavius 4D -PTW dosimetry that incorporates 729 chamber array. The gamma analysis passing rates were calculated by verisoft software using criteria of 3% for dose and 3 mm for distance to agreement and based on the global maximum dose of calculation volume and suppressing the dose blow of 5% of maximum dose of the calculated volume. Then we tested the correlation between the QA passing rate with the related MLC matching percentage. The gamma analysis passing rate was in the range of 90%–99.1% and the matching percentage ranged from 89% to 98.8%. The assessment of the correlation between the gamma passing rate and the matching percentage showed 0.45 correlation coefficient. The p-value was less than 0.05 which indicates a statistically significant correlation. The effect size was on the order of 0.7. Trajectory log file analysis combined with the measurement-based patient specific QA is beneficial to our IMRT QA process and it could be an aid to detect the actual delivery error and hence could improve our passing rates in the future.