Click here to

Session: Multi-Disciplinary General ePoster Viewing [Return to Session]

A New Metric of Contour and Dose Accuracy Applied to Automated Adaptation Features in a Commercial TPS

M Bogue1*, J Chen2, E Yu3, J Patrick4, R Munbodh3, (1) University of Rhode Island, Kingston, RI, (2) Brown University, Providence, RI, (3) Alpert Medical School of Brown University/Rhode Island Hospital, Providence, RI, (4) London Regional Cancer Program, London, ON, CA


PO-GePV-M-107 (Sunday, 7/10/2022)   [Eastern Time (GMT-4)]

ePoster Forums

Purpose: Adaptive radiation therapy (ART) uses images acquired at regular intervals during treatment to modify the treatment plan in response to observed anatomical changes. Cone-beam computed tomography images (CBCTs) acquired on the treatment unit can be used for this purpose. Previous studies assess the accuracy of these techniques by comparing automated adaptive contouring and dosimetry to contours and doses computed on conventional CT according to standard metrics, such as comparing the DVH curve, contour overlap, and gamma analysis of the resulting dose distribution. These metrics compare contours or doses, but not both. In particular, the accuracy of CBCT images is typically higher near the isocenter, which is also usually near the high-dose region, but contour agreement metrics do not account for this. We introduce the dose-volume agreement (DVA) curve for comparing two contours relative to associated dose distributions, and apply it to the evaluation of contour deformation and CBCT dose computation in RayStation.

Methods: We identified 19 patients who had received radiotherapy of the head and neck with periodic CBCT imaging for setup verification and a rescan after several weeks. CBCT densities were assigned using RayStation's HU binning calibration or deformable image registration (DIR) of the planning CT to the CBCTs. Contours were computed automatically by DIR, and raw automated contours were adjusted by a physician. Automated and adjusted CBCT contours and doses computed from each density profile were compared to doses and contours on the rescan CT.

Results: DVA curve points were found to reveal improvements in both contouring and dose accuracy, specifically highlighting contour disagreements in high-dose regions.

Conclusion: Progress continues to be made towards the general adoption of adaptive radiotherapy. Increasingly detailed metrics of dosimetric accuracy will help guide the development of automated systems that support the high workload inherent to adaptive dose tracking and replanning.


Cone-beam CT, Image Correlation


Not Applicable / None Entered.

Contact Email