Author: Sina Mossahebi, Pouya Sabouri, Kayla Schneider, James W Snider 👨🔬
Affiliation: University of Maryland School of Medicine, Proton International, Department of Radiation Oncology, University of Arkansas for Medical Sciences (UAMS), Department of Radiation Oncology, University of Arkansas for Medical Sciences 🌍
Purpose:
Conventional patient-specific QA (PSQA) for intensity-modulated proton therapy (IMPT) requires extensive measurements, straining resources in single-room proton centers. This study evaluates the feasibility of replacing measurement-based PSQA with a log file-based computational approach using commercial software. We assess whether independent dose recalculations from machine log files can improve PSQA efficiency while maintaining treatment accuracy.
Methods:
PSQA measurements from 77 patient plans treated at a single-room IBA ProteusOne facility were retrospectively analyzed. Clinical plans were optimized and calculated using the Monte Carlo algorithm (RaySearch, Sweden). PSQA was performed at two to three depths using a 3%/3 mm gamma analysis. Log files from the first treatment session were processed in myQA iON (IBA Dosimetry, Germany) to recompute the delivered dose, which was then compared to the TPS-calculated dose. The lowest gamma pass rate among the measured depths was compared to the external contour gamma pass rate from myQA iON. Statistical analysis, including paired t-tests, Bland-Altman analysis, and least squares regression, was conducted to assess agreement, trends, and correlations between measurement-based and log-based QA.
Results:
The minimum/mean pass rates for measurement-based QA were 95.1%/99.35%, while for log-based myQA iON, they were 88.4%/98.37%. A paired t-test (T=7.12, p<1.18×10⁻¹¹) showed a significant difference, with log-based QA yielding lower pass rates, suggesting higher sensitivity than measurement-based PSQA. The mean difference between methods was 0.98%, indicating minor underestimation. Bland-Altman analysis showed greater discrepancies at lower pass rates, while regression analysis (R² = 0.001, p = 0.628) suggested variability across treatment sites.
Conclusion:
Log file-based PSQA provides a viable alternative to conventional measurement-based PSQA for IMPT. While a statistically significant difference was observed between the two methods, the mean difference of 0.98% suggests that the log-based approach systematically yields slightly lower pass rates, likely reflecting a stricter PSQA threshold rather than an inherent limitation of the method.