S2382
Physics - Quality assurance and auditing
ESTRO 2026
unique). Results varied significantly, even between submissions confirmed to have matching protocol. In the most extreme case, a single institute with multiple sets of submitted results achieved a sensitivity of 0.0 (passed all plans that ‘should fail’) and a sensitivity of 1.0 with the same PSQA device and protocol on beam- matched linacs. A ‘perfect’ result (sensitivity and specificity both equal to 1.0) was also achieved using a range of gamma criteria for matching devices. All commonly used devices ( ≥ 5 submissions) had at least one submission with optimal results (Figure 2).
Proffered Paper 3536
It's not your device, it's how you use it: sensitivity and specificity of PSQA to simulated linac errors in a large multi-centre study Lauren May 1 , José Antonio Baeza-Ortega 1 , Mohammad Hussein 2 , Sarah Porter 3 , Alisha Moore 3 , Michael Barnes 4 , Peter B Greer 1,4 , Catharine H Clark 5,6 , Joerg Lehmann 4,1 1 School of Information and Physical Sciences, University of Newcastle, Newcastle, Australia. 2 Department of Radiation Oncology, King Khalid University Medical City, Abha, Saudi Arabia. 3 Radiation Therapy Quality Assurance, Trans Tasman Radiation Oncology Group (TROG Cancer Research), Newcastle, Australia. 4 Department of Radiation Oncology, Calvary Mater Hospital, Newcastle, Australia. 5 Metrology for Medical Physics Centre, National Physical Laboratory, Teddington, United Kingdom. 6 Department of Medical Physics and Bioengineering, University College London, London, United Kingdom Purpose/Objective: Despite published guidelines and recommendations [1], there is a large diversity in the clinical implementation of PSQA [2, 3]. The SEAFARER study remotely assesses PSQA systems’ sensitivity and specificity to deliberately introduced delivery errors [4]. This work aimed to identify trends in PSQA device protocol (gamma criteria and other settings/methodology) and performance. Material/Methods: SEAFARER PSQA audits for a head and neck (H&N) case and stereotactic ablative radiotherapy spine case have been undertaken. For each case, baseline plans meeting predefined clinical goals were created for common linac models. Modified copies of the baseline plans with introduced errors in MLC position, collimator and gantry angle, and machine output were created (11 plans for H&N, 12 plans for spine plus 10 optional). Participants were asked to perform their clinical PSQA on each modified plan with reference back to the baseline plan, without knowing the modifications included, and to report whether each plan ‘passed’ or ‘failed’. Details of PSQA protocol used were collected as optional open-ended responses. Sensitivity and specificity analyses were conducted with plans which ‘should fail’ defined as having an increase in OAR dose or target coverage reduction of >5%. Results: Across both studies, 19 different PSQA devices were used and three combinations of 2–4 devices (Figure 1). Variously detailed descriptions of PSQA protocols were provided for 83 of 90 total sets of PSQA results submitted for the H&N plans (at least 53 being unique) and 54 of 67 total spine submissions (at least 34
Figure 1. PSQA devices used.
Figure 2. Heatmaps showing sensitivity and specificity of individual submissions for two commonly used devices for each study. Conclusion: All the commonly used PSQA devices are capable of obtaining a high level of sensitivity and specificity. Differences in PSQA performance go beyond device and protocol used, potentially affected by compounding factors, therefore protocols should be developed and tested per department.
Made with FlippingBook - Share PDF online