Program Information
Knowledge-Based Automated Planning for Oropharyngeal Cancer
A Babier1*, J J Boutilier1, A L McNiven2, T C Y Chan1, (1) University of Toronto, Toronto, Ontario, (2) Princess Margaret Hospital, Toronto, ON
Presentations
TH-CD-205-5 (Thursday, August 3, 2017) 10:00 AM - 12:00 PM Room: 205
Purpose: To automatically generate intensity-modulated radiation therapy plans that match or surpass clinical oropharynx plans, by combining knowledge-based planning (KBP) predictions with an inverse optimization (IO) pipeline.
Methods: We generalized a prior KBP model that used overlap volume histograms to predict achievable dose volume histograms (DVHs). Using leave-one-out cross-validation, we applied this method to a large dataset of 217 oropharynx patients. The predicted DVHs were input into an IO pipeline that generated treatment plans (KBP plans) via an intermediate step using estimated objective function weights and an inverse planning model. KBP plan DVHs were compared to the predicted DVHs and clinical plan DVHs to assess the KBP-IO pipeline. To isolate the effect of the KBP predictions, we also put clinical DVHs through the IO pipeline to produce clinical inversely optimized (CIO) plans. The KBP plans were benchmarked against the CIO plans using DVH differences and clinical planning criteria.
Results: Compared to clinical plans, KBP plans consistently achieved lower dose to OARs (5.3Gy median reduction), with the largest median reduction to the esophagus (10.6Gy) and the smallest median reduction to the spinal cord (4.1Gy). Clinical planning criteria for OARs such as the larynx were satisfied more frequently by KBP plans (80%) than for the CIO (61%) and clinical (60%) plans. The KBP plans also satisfied 93% of planning criteria for the high-dose targets, compared to the CIO (86%) and clinical (89%) plans. However, KBP plans satisfied criteria for low-dose targets at a lower rate (38%) compared to the CIO (50%) and clinical (55%) plans.
Conclusion: Our automatically generated KBP plans can replicate, and typically improve upon, the dose to OARs and primary target coverage observed in clinical treatment plans for a very large cohort of oropharynx patients.
Contact Email: