Deep-Learning-Based Dose Predictor for Glioblastoma-Assessing the Sensitivity and Robustness for Dose Awareness in Contouring

Simple Summary For accurate radiotherapy, a clear definition of the geometric extent of organs and tumor volumes is important. Due to the laborious task of manually drawing contours to define these, automatic segmentation models are becoming increasingly available. These models, however, need to be visually evaluated by radiation oncology experts. This evaluation itself takes up valuable time, therefore making an efficient and clinically relevant validation of auto-segmented results desirable. An accurate 3D dose prediction model can help create dose awareness prior to the actual dose-planning step. It can provide useful information for the quality assurance of the contouring step. In this study, we trained a 3D dose predictor for volumetric modulated arc therapy (VMAT) treatment of glioblastoma patients based on an existing architecture called a cascaded 3D U-Net. We further tested this model's sensitivity and robustness for the purpose of estimating dose changes due to contour variations.Abstract External beam radiation therapy requires a sophisticated and laborious planning procedure. To improve the efficiency and quality of this procedure, machine-learning models that predict these dose distributions were introduced. The most recent dose prediction models are based on deep-learning architectures called 3D U-Nets that give good approximations of the dose in 3D almost instantly. Our purpose was to train such a 3D dose prediction model for glioblastoma VMAT treatment and test its robustness and sensitivity for the purpose of quality assurance of automatic contouring. From a cohort of 125 glioblastoma (GBM) patients, VMAT plans were created according to a clinical protocol. The initial model was trained on a cascaded 3D U-Net. A total of 60 cases were used for training, 15 for validation and 20 for testing. The prediction model was tested for sensitivity to dose changes when subject to realistic contour variations. Additionally, the model was tested for robustness by exposing it to a worst-case test set containing out-of-distribution cases. The initially trained prediction model had a dose score of 0.94 Gy and a mean DVH (dose volume histograms) score for all structures of 1.95 Gy. In terms of sensitivity, the model was able to predict the dose changes that occurred due to the contour variations with a mean error of 1.38 Gy. We obtained a 3D VMAT dose prediction model for GBM with limited data, providing good sensitivity to realistic contour variations. We tested and improved the model's robustness by targeted updates to the training set, making it a useful technique for introducing dose awareness in the contouring evaluation and quality assurance process.

Automated summary

Accurate radiotherapy requires clear definition of organs and tumor volumes, and automatic segmentation models can improve efficiency. In this study, a 3D dose prediction model was trained for glioblastoma radiation therapy and tested for sensitivity and robustness to contour variations.