Program Information
A Novel Automated Inverse Planning Optimization Using Organ Specific Line of Defense (OSLD) Model
H Zhang1*, S Huh2 , M Reilly3 , T Zhao4 , S Olberg5 , D Yang6 , J Kim7 , S Mutic8 , J Park9 , (1) Washington University School of Medicine in St. louis, Saint Louis, MO, (2) University of Florida Proton Therapy Institute, Jacksonville, Florida, (3) Washington University School of Medicine, St. Louis, MO, (4) Washington University School of Medicine, St. Louis, MO, (5) Missouri University of Science and Technology, Rolla, Missouri, (6) Washington University in St Louis, St Louis, MO, (7) Yonsei University College of Medicine, Yonsei Cancer Center, Seoul, ,(8) Washington University in St Louis, St Louis, MO, (9) Washington University in St. Louis, St. Louis, MO
Presentations
MO-F-CAMPUS-TT-1 (Monday, July 31, 2017) 4:30 PM - 5:30 PM Room: Therapy ePoster Theater
Purpose: To provide a clinically acceptable set of optimized radiotherapy treatment plans while minimizing the complexity of objective functions and trials-and-errors to facilitate the IMRT/VMAT treatment planning process.
Methods: The proposed method constructs the plan objective functions based on the organ specific dose limiting structures (OSLD) that are modeled by an analysis of the geometric arrangements of planning target volume(s) (PTV) relative to the neighboring OARs and the dose distribution of the most conformal plan. Initially, a tightest dose distribution D_tight is computed by optimizing sole objective functions on PTV(s). Secondly, the spatial dose distribution and dose volume histogram (DVH) are evaluated to estimate the maximum dose reduction to each OAR at selected dose d_i. At this stage, machine learning or a knowledge-based technique (e.g. Rapid PlanTM) can be utilized. Thirdly, the OSLDs are modeled by deforming the isodose lines d_i such that the region(s) intersecting OARs shrink towards the PTV and the others expand to preserve initial volume of d_i, preventing the “dose splash” effect. To construct the OSLDs, we applied a modified active contour model with balloon force. Finally, the objective function is constructed with a set of PTV(s) and OSLDs. In this manner, the overall complexity of the planning objective function is reduced significantly.
Results: The proposed method was tested on clinically treated prostate and head and neck patients. The plan quality was comparable to the clinically approved plans. The proposed plan used 8 vs 14 objectives in the prostate and 10 vs 92 in the H&N case. Both plans were achieved in two trials (D_tight + Final).
Conclusion: This work demonstrates a novel automated treatment planning that greatly reduces the number of objective functions and trials without compromising the plan quality of conventional treatment planning. We anticipate that the proposed method will significantly facilitate treatment planning.
Contact Email: