Program Information
Framework for the Quantitative Assessment of Clinical Adaptive Radiation Therapy Protocols
S Weppler*, H Quon , R Banerjee , C Schinkel , W Smith , Tom Baker Cancer Centre, Calgary, AB
Presentations
SU-E-FS4-3 (Sunday, July 30, 2017) 1:00 PM - 1:55 PM Room: Four Seasons 4
Purpose: Although adaptive radiation therapy (ART) has been shown to improve clinical outcome, it is unclear which “flagging” metrics (i.e., change in body contour, relative weight loss) most effectively identify patients in need of replanning. Conventionally, ART flags are used to pre-screen replan candidates and final decisions regarding patient-specific benefits of replanning are subsequently made by treating clinicians. We propose a framework to allow institutions to assess ART flagging efficacy with respect to candidate identification and resource allocation.
Methods: Quantitative evaluation of a flagging metric requires identification of unacceptable deviations in dose parameters (e.g., we surveyed treating physicians). Periodically acquired cone-beam CT images of a retrospective cohort are used to deform copies of the CT-simulation permitting contour propagation and dose recalculation. Survey data serve as “ground truth” criteria in truth table analyses. A meaningful flag requires high sensitivity (>90%) so that patients in need of a replan are identified; reduction of the false positive fraction improves the use of clinical resources. This framework has been implemented on a retrospective cohort of fifteen head and neck cancer patients flagged for potential replan at our center by exhibiting a change in external body contour exceeding 1.5 cm.
Results: By the proposed framework, the body contour-based flag was found to have poor sensitivity due to the discrepancy between the protocol’s intention (identify increases/decreases in dose) and the parameters most of interest to physicians (spatial divergence of target region and high dose volumes). During clinical implementation, the protocol also erroneously flagged setup uncertainties in shoulder position for ten of fifteen patients.
Conclusion: We identified inconsistencies in the intention, clinical relevance, and implementation of the body contour-based flag. This framework provides quantitative assessment and optimization of ART protocols for meaningful improvements in replan candidate identification and resource allocation.
Contact Email: