Quantifying Adaptive Triggers Using Geometric, Dosimetric, and Radiobiological Metrics
Abstract
Purpose
Daily anatomical changes can invalidate the geometric, dosimetric, and biological assumptions of reference treatment plans. Quantitative evidence describing fraction-level deviations across disease sites remains limited, and adaptive workflows are often driven by qualitative assessment or fixed mid-course replanning. This study uses daily delivered dose metrics to characterize temporal behavior in Head & Neck (H&N) and Prostate SBRT and to inform site-specific adaptive strategies.
Methods
A web-based Dashboard computed five metrics per fraction: interfraction Dice, intrafraction Dice, PTV and OAR objective scores, a radiobiology score, and a composite radar area. Seventy-two patients were analyzed: H&N (n=18, 28–35 fractions, ~70 Gy) and Prostate SBRT (n=54, 36.25 Gy in 5 fractions on MR-Linac). Inclusion required reference plan, structure set, and dose DICOMs. Daily RT Dose files were imported into a PostgreSQL backend, processed by the Dashboard, and evaluated across fractions to identify temporal trends and fraction-specific outliers.
Results
H&N treatments exhibited clear time-dependent degradation. Interfraction Dice worsened with week of treatment, and OAR objective and radiobiology scores increased with broadening distributions. Radar area expanded over time, indicating progressive global deterioration relative to the reference plan, while intrafraction Dice remained low and stable. PTV objective scores were elevated throughout, highlighting persistent sensitivity to small geometric deviations. In contrast, Prostate SBRT showed substantial fraction-to-fraction variability without temporal drift. Interfraction Dice, PTV objective, and radiobiology scores were frequently elevated from the first fraction, but acceptable fractions also occurred, underscoring patient- and day-specific behavior.
Conclusion
H&N behaves as a time-dependent system with cumulative degradation, supporting threshold-based mid-course reassessment. Prostate SBRT demonstrates persistent, fraction-specific variability, supporting daily plan evaluation rather than fixed replanning schedules. Metric-based decision support may enable site-specific adaptive policies that improve resource allocation and better reflect clinical reality.