Evaluation of AI-Driven Adaptive Workflow for High Quality Palliative Radiotherapy
A Kejda1, S Wong1, S Roderick1, J Booth1,2*, T Eade1,3, (1) Northern Sydney Cancer Centre, Royal North Shore Hospital, Sydney, Australia, (2) University of Sydney, Sydney, New South Wales, Australia, (3) Sydney Medical School, University of Sydney, Sydney New South Wales, Australia,
Presentations
PO-GePV-M-82 (Sunday, 7/25/2021) [Eastern Time (GMT-4)]
Purpose: Approximately one third of radiotherapy patients are treated with palliative intent. Varian’s Ethos online adaption system utilises AI contouring on daily cone-beam CT (CBCT) to auto-generate a plan of the day. This work evaluated the Ethos adaptive workflow in palliative patients comparing clinician contours against Ethos AI segmentation, and corresponding adaptive plans against original (scheduled) plans.
Methods: Ten palliative patients (27 fractions) were replanned with dynamic MLCs using Ethos TPS, then duplicated and sorted into two test arms. Adaptive treatment was simulated in a test environment using patient-specific CBCTs. In arm 1, clinicians edited all AI contours before an adaptive auto-plan was generated. In arm 2, no contour edits were made. The adaptive plans generated were compared using metrics assessed in original plans (PTV D95% and 100% conformity index). On a subset of six fractions, AI and clinician-modified structures were compared using the DICE similarity coefficient.
Results: All adaptive plans passed the primary constraint PTV D95%>95%, while two scheduled fractions failed. Mean PTV D95% for adaptive and scheduled fractions were 102.4% and 104.0%, respectively, with no statistically significant difference found (t-test P=0.29). Mean conformity indices measured 0.97 and 0.85 for adaptive and scheduled plans, respectively, were found to be statistically different (t-test P=0.017). The DCEs for CTV, kidneys, liver, small bowel and stomach were 0.90±0.09, 0.930±0.05, 0.98±0.02, 0.928±0.04 and 0.70±0.11, respectively, indicating accurate AI segmentation of all contours except for the stomach.
Conclusion: Ethos AI-driven adaption was shown to achieve better target coverage than scheduled plans, indicating the system could account for inter-fraction anatomy changes associated with palliative cohorts. Good agreement was shown between preliminary Ethos and clinician contours, except for the stomach, which was most likely impacted by the presence of gas causing image artefacts, impacting AI segmentation. Future work includes online adaption in diagnostic CT-enabled palliative planning.
Funding Support, Disclosures, and Conflict of Interest: Northern Sydney Cancer Centre has a funded collaborative research agreement with Varian
ePosters
Keywords
Not Applicable / None Entered.
Taxonomy
TH- External Beam- Photons: adaptive therapy
Contact Email