Progressive Knowledge Modeling for Pelvic IMRT/VMAT Treatment Planning

Intensity Modulated Radiation Therapy (IMRT) and Volumetric Modulated Arc Therapy (VMAT) have become effective tools for treating cancer with radiation. Designing a high quality IMRT/VMAT treatment plan is time consuming. Different knowledge-based methods are being developed to reduce planning time and improve plan quality by extracting parameters from prior expert plans to form knowledge models and applying such models to the new patient cases. Currently, these methods mostly generate models for one particular cancer type and therefore various disease types require training of multiple knowledge models with a large number of cases.;To investigate the feasibility of IMAT/VMAT treatment planning knowledge modeling for multiple cancer types, a progressive study was designed to build a treatment planning knowledge model that quantifies correlations between patient pelvic anatomical features and the Organ-At-Risk (OAR) sparing features from different disease types. Low-risk prostate plans with relatively simple Planning Target Volume (PTV) to OAR geometry, which has been the most common geometry type studied in previous knowledge based studies, were used to train the model as the starting point of the progressive modeling process. Cases with more complex PTV-OAR anatomies (prostate cancer cases with lymph node irradiation and anal rectal cancer cases) were added to the training dataset sequentially until the model prediction accuracies reached plateau. The Dose Volume Histograms (DVHs) predicted by the knowledge model for the bladder, the femoral heads and the rectum were validated by clinical plans from all three types of cases. Dosimetric parameters extracted from the predicted DVHs and the corresponding actual plan values were compared for prediction accuracy of this multi-disease type knowledge model. Further, the prediction accuracy was also compared with the models trained with three single disease type cases (including low-risk prostate cancer (type 1), high-risk prostate cancer with lymph nodes (type 2) and anal rectal cancer (type 3), respectively).;Prediction accuracy reached a plateau when six high-risk prostate cancer cases and eight anal rectal cancer cases were added to the training dataset. The determination coefficients R2 for the OARs are: bladder: 0.90, rectum: 0.64 and femoral heads: 0.82. There is no significant difference in prediction accuracy between the multi-disease type model and the single-disease type models (F-test p-value: bladder: 0.58, rectum: 0.97 and femoral heads: 0.44).