Session Item

The recent introduction of the log files during patient specific QA requires a deep analysis process in order to integrate them into clinical QA workflow. We investigated the error sensitivity of log files analysis using LinacWatch^® (LW) and compared it with the one determined using a pre-treatment QA phantom (Octavius4D, Oc4D).

A total of 12 clinical plans have been considered: 3 prostate, 3 rectum, 3 head and neck (H&N) and 3 breast. In order to analyze error detection sensitivity, for each of them, new plans were created introducing a set of errors regarding MLC bank shift, jaws deviation, collimator and gantry rotation and monitor unit (MU) over dosage for a total of 180 plans. They were delivered on an Elekta Synergy linac. Fluence maps both for Oc4D and LW were simultaneously acquired. The comparison between error-free plan and plan with intended errors (both using Oc4D and LW) was performed with local γ-analysis. Personalized protocols were chosen: for Oc4D,  3%/3 mm for H&N and 2.5%/2.5 mm for the others; for LW, 2.5%/2.5 mm for H&N and 2%/2 mm for the others. Deviations of the collimator and gantry errors in LW were performed using the RMS evaluation.

The total sensitivity was 32.1% for Oc4D and 36.9% for LW. For MLC errors, small deviation of ±0.25 mm produced a sensitivity of 8.3% for Oc4D vs 0% for LW. Both the systems were able to detect an MLC shift of ±1 mm with the same sensitivity (58.3%). Jaw deviations of ±0.5 mm had no effect on sensitivity for both systems. LW detected 70.9% and 87.5% collimator and gantry errors compared to 41.7% and 66.7% of Oc4D. The sensitivity of MU errors was 12.5% for Oc4D vs 37.5% for LW. If we lower the γ-criteria in LW for prostate and breast treatments to 1.5%/1.5 mm while maintaining the specificity always at 100%, the sensitivity in detecting small MLC deviations increased from 0% to 8.3% and from 8.3% to 25% for ±0.25 mm and ±0.5 mm errors, respectively. 

Our result in the context of the patient specific quality assurance of all VMAT and IMRT plans, indicate that the log file analysis with LinacWatch can be considered as a complementary system to Octavius4D phantom, in particular to control each treatment session. Anyway, to increase LW sensitivity, we suggest using personalized protocols in the γ-analysis, depending on different treatment sites.