Cohortdvh: Plan-Level Contextualization of Radiation Therapy Quality
Abstract
Purpose
As treatment planning evolves toward patient-specific optimization, traditional static dose constraints are insufficient for distinguishing high-quality versus acceptable plans. Current planning directives rely on generic dose constraints that fail to strive for clinically optimized plans. We present CohortDVH, a DVH analytics tool that aggregates a clinic’s historical treatments to contextualize individual plans, providing a visual framework to assess plan quality relative to institutional standards.
Methods
We compiled a comprehensive database of all treated cases by ingesting ~35,000 archived DICOM plans spanning multiple treatment planning systems. Prescriptions were resolved through a hybrid DICOM-OIS pipeline to ensure accurate cohort assembly despite incomplete or system-specific metadata. We developed CohortDVH for plan evaluation, overlaying patient-specific DVHs against population-derived DVH summaries (median, inter-quartile range, and 10th-90th percentile band) generated from matched cohorts, allowing users to evaluate plan quality relative to historic precedent.
Results
CohortDVH serves multiple functions including: -Clinical Decision Support: Instantly benchmark a specific plan against a cohort of treated patients, identifying outlier plans that may require re-optimization. -Education: Visualize the range of expected quality to provide feedback for trainees to validate practice plans and internalize the practical achievable dosimetry. -Longitudinal QA: Tracking plan quality over time to quantify the impact of technological advancements on measurable population-level improvements. -Protocol Refinement: Providing empirical data for continuous improvement and optimization of planning constraints. By evaluating current institutional constraints compared to achieved population DVHs, goal constraints can be continuously refined, with more stringent goals where appropriate and relaxing unrealistic constraints, thus closing the loop between clinical practice and planning policy.
Conclusion
CohortDVH bridges the gap between individual plan geometry and historical clinical consensus. Visual outlier detection with comprehensive population data supports transitioning practice from simply meeting single value acceptable dose constraints to maximizing achievability, ensuring plan quality is continuously assessed and improved in the modern clinic.