Program Information
Fully Automated VMAT Planning in the Head and Neck
R McCarroll1*, L Zhang1 , C Cardenas1 , P Balter1 , K Kisling1 , M Mejia2 , C Nelson1 , D Followill1 , C Peterson1 , J Yang1 , B Beadle3 , L Court1 , (1) UT MD Anderson Cancer Center, Houston, TX,(2) University of Santo Tomas Hospital - Benavides Cancer Institute, Manila, Metro Manila, (3) Stanford University, Palo Alto, California
Presentations
TH-CD-205-3 (Thursday, August 3, 2017) 10:00 AM - 12:00 PM Room: 205
Purpose: To fully automate 2arc VMAT plans for head-and-neck cancer, including isocenter detection, body detection, automated contouring of normal tissues and intermediate- and low-risk CTVs (2 and 3), jaw settings, and inverse planning.
Methods: In-house algorithms were developed to contour the body structure, remove the CT table, detect treatment isocenter, set jaws, set collimator angle, and send and receive secondary dose calculation metrics. In-house multi-atlas auto-contouring was used to contour normal tissues and intermediate- and low-risk CTVs. Physician drawn GTVs were used to initiate creation of high risk CTVs. Planning PTVs were created using 5mm margins. Through the use of Application Programming Interface (ESAPI), Eclipse RapidPlan was used to automatically create 2arc VMAT plans. DVH analysis of the auto-plan was compared to the clinically delivered plan. To assess dose delivered to true normal tissues, the auto-plan was applied to physician-drawn normal structures and DVH metrics assessed.
Results: For 30 plans, average plan time (no intervention required) 47.9 minutes, not including GTV creation. Average 95% prescription coverage of high-, intermediate-, and low-risk PTVs were 97.9%, 98.5% and 98.8%, respectively[PTV1 range: 95.4-99.8%]. For all plans, auto-contoured and physician drawn spinal cord and brainstem met maximum dose constraints [Dmax<45,50Gy]. For 27/28 plans (excluding 2 bilateral disease cases), contralateral parotid met constraints of either Dmean<26Gy (16/28) or V50<30Gy (27/28 patients). For 5 plans, brain constraint (Dmax<54Gy) was not met, however, in 4 of these cases the clinical plan was not able to meet this constraint either. Fifteen plans (dose distributions and contours) were reviewed by a head-and-neck radiation oncologist, of which 14 were acceptable for treatment (without edit of contours, beams, etc.).
Conclusion: Combining in-house and commercial software, we have developed a workflow which fully automates the treatment planning process for head-and-neck VMAT plans. Tested on 15 patients, 93% were rated acceptable for planning.
Funding Support, Disclosures, and Conflict of Interest: This work is funded by the NIH (UH2CA202665). Additional support from Varian Medical Systems and Mobius Medical Systems.
Contact Email: