Program Information
Stochastic Frontier Analysis Used for Rectum Sparing in VMAT Planning for Prostate Cancer
A Kroshko1*, N Varfalvy1 , O Morin2 , L Archambault1 , (1) CHUQ Pavillon Hotel-Dieu de Quebec, Quebec, Qc, (2) University of California San Francisco, San Francisco, CA
Presentations
TU-D-108-4 (Tuesday, August 1, 2017) 11:00 AM - 12:15 PM Room: 108
Purpose: Radiation Therapy treatment planning for advanced technique such as Volumetric Modulated Arc Therapy (VMAT) often results in a compromise between delivering the prescribed dose to the target and sparing organs-at-risk (OAR). We aim to develop metrics based on patient specific morphology to improve rectal sparing in prostate cancers treated with VMAT. By using Stochastic Frontier Analysis (SFA), it is possible to determine the lowest achievable rectal dose in terms of geometrical features.
Methods: A retrospective study of 310 patients treated for prostate cancer was made. Patients were divided in 3 groups of about a hundred patients each based on their prescribed dose. Geometric parameters such as overlap volume, overlap gradient and average Hausdorff distance were extracted using python scripts within the 3D Slicer platform. Relationships between geometry of the rectum and dosimetric parameters were defined by using SFA. In this method, the distribution of the treatment plans is supposed asymmetric with respect to an optimal frontier. A maximum likelihood technique is used to determine the distribution and the frontier parameters.
Results: SFA optimal frontiers were obtained for several dosimetric parameters such as V40Gy, V50Gy and V65Gy where more than 80% of the plans could be improved. At least two geometric parameters were used to obtain a frontier. These frontiers represent an achievable minimum dose for a given linear combination of significant geometric parameters determined by their foremost correlation with a dosimetric parameter.
Conclusion: We were able to obtain SFA frontiers for VMAT treatment plans for prostate cancer. By using the geometric and dosimetric information of previously treated patients, with SFA, it is possible to predict the achievable sparing of OAR for future plans depending on their specific morphology. With these metrics, we wish to develop a tool able to help the planning of advanced therapy technique such as VMAT.
Contact Email: