Session Item

Being a busy department with 5700 new patient plans/year and several satellite centers, efficiency, standardization and automation are key for a QA program. A web-based system was installed early 2017, for pre-treatment and in-vivo QA based on phantomless EPID and log files (PerFRACTION^TM, Sun Nuclear Corporation). Images and log files are actively retrieved. Calculation and analysis occur automatically in the background. A clinical validation of the system’s performance on detection of errors and a comparison with the 1^st year results will be reported, along with an analysis of causes and actions taken for failed fractions.

In this study results are reported for all patients treated between Sept 2018 and Aug 2019, compared with the results from the 1^st year (Oct 2017 to Aug 2018). For all patients receiving photon treatment, logfiles were automatically analyzed. For all VMAT plans, pre-treatment QA was performed (‘Fraction 0’). Transit EPID images were generated the first 3 days of treatment and weekly thereafter. In the 1^st year these had been compared using relative (image to image) analysis. In the 2^nd year absolute verification was introduced allowing comparison of the images to calculated data, enhancing detectable errors.

56542 fractions were analyzed: 91% passed, 7% failed and 2% were not calculated. As in the 1^st year, no relevant patient or machine errors were detected with analysis of log files alone. Pre-treatment QA showed results comparable to the 1^st year, with clearly better results for TrueBeam than for Clinac (Fig 1). In 43% of the treatment fractions, transit EPID dosimetry was performed. 16% of these fractions failed, compared to 15% in the 1^st year. When analyzing the 3619 failed fractions (FF) of non-stereotactic patients, the number of false positives has slightly decreased from 41% to 38%, with much better results for TrueBeam (19%) than for Clinac (49%), Fig 2. The number of FF due to imager position or calibration has decreased from 31% to 17%, but the number of FF due to technical imager or machine problems increased from 8% to 18%. The number of FF due to problems with patient positioning decreased from 40% to 32%. Those due to patient anatomy increased from 19% to 28%. Actions for FF were comparable to the 1^st year with an increase of the number of plan adjustments from 1% to 4%. Because of the introduction of absolute verification more errors were caught such as: weight loss at start of treatment, problem with bellyboard, errors in planning, problems at simulation with 4DCT artefacts or contrast agents in bowel, pleural effusions cleared up by the time of treatment, poor breathing for gated breast patients…

Absolute verification for transit in vivo dosimetry enhanced detectable errors. The number of false positives was clearly lower for TrueBeam than for Clinac. The number of plan adjustments increased, showing the increased confidence in the system as a base for adaptive planning.