ESTRO 2024 - Abstract Book

S4746

Physics - Quality assurance and auditing

ESTRO 2024

Radiobiological dose corrections had a larger dosimetric impact than the DIR algorithm uncertainties among the participating centers, even in this difficult head and neck case. More clinical cases and comprehensive evaluation of relevant OARs are required to strengthen this conclusion.

Keywords: re-irradiation, dose accumulation, head and neck

267

Digital Poster

An in-house machine log-based tool for monitoring MLC positional errors in online treatment

Pak Hang Nam, Bin Yang, Max W.K. Law, Kin Yin Cheung, Siu Ki Yu

Hong Kong Sanatorium & Hospital, Medical Physics Department, Hong Kong, Hong Kong

Purpose/Objective:

A multitude of methods have been employed to ensure the quality of patient’s treatment, particularly when it comes to patient-specific quality assurance (QA). Among these methods, measurement-based verification QA stands as a more comprehensive means of examining treatment plan excellence compared to calculation-based verification QA. However, this method demands a considerable time investment. In light of the burgeoning popularity of online treatment monitoring and online adaptive treatment, there arises a need for QA analysis tools that leverage the machine log collected during treatment. This study aims to develop a machine-log-based QA tool, which is efficient and accurate for streamlining the QA workflow in MR-Linac treatments.

Material/Methods:

All measurements in this study were conducted on a commercial 1.5T MR-Linac. Following the delivery of each group of fields, a machine log was automatically generated, documenting the machine's status at an interval of 0.04 seconds throughout the plan delivery. The log encompassed various data, including but not limited to (1) positional error of each leaf(mm), (2) precise position of each leaf(mm), and (3) time stamp(ms). Our in-house tool, which was developed using MATLAB, was capable of conducting statistical analysis of the data extracted from the log. This analytical process yielded multiple indexing results, including but not limited to (1) mean leaf error (mm), and (2) leaf error events that surpassed the pre-set tolerance (%).

Performance of this QA tool was evaluated in several aspects.

Firstly, we compared our QA tool results to the routinely performed multi-leaf collimator (MLC) Picket Fence (PF) test results to evaluate their correlations and further to investigate if the MLC performance would foresee the occurrence of an MLC machine interlock event based on the log.