Error detection sensitivity and operational efficiency of phantom-based and phantom-less patient-specific quality assurance in proton therapy.

Abstract

Background and purpose: In proton therapy (PT), patient-specific quality assurance (PSQA) is an important component of the measures which ensure accurate and safe treatment delivery. Traditional phantom-based PSQA is resource-intensive and may miss clinically relevant data transfer and delivery errors. This study compared the sensitivity and operational effort of established phantom-based methods versus an automated, phantom-less alternative based on the increasingly utilized log file-based approach.

Materials and methods: We evaluated phantom-based (specifically including dose measurements & manual physics checks) and phantom-less (specifically based on log file-based QA & automated physics checks) PSQA workflows. Twenty-nine artificial error scenarios were introduced to a clinically delivered head-and-neck plan. Error detection sensitivity was determined via the percentage of scenarios detected by each workflow. Operational effort was quantified by counting mouse clicks and manual parameter entries in clinical systems.

Results: Phantom-less PSQA detected 90% of simulated errors, outperforming phantom-based PSQA detecting only 52% at clinically realistic measurement conditions. Specifically, log file-based QA alone detected 83% of scenarios versus 10% detected by phantom-based measurements. Manual and automated plan parameter checks showed an equal sensitivity, detecting 48% of scenarios. The more automated, phantom-less alternative could reduce operational effort by at least one third of the currently required mouse clicks.

Conclusions: Phantom-less PSQA based on log file-based QA and automated physics checks provided higher error detection sensitivity and significantly reduced manual operational effort compared to conventional phantom-based and manual methods. These findings support its integration into clinical practice, a key objective of many PT centers.