Deep learning models, such as U-Net, can be used to efficiently predict the optimal dose distribution in radiotherapy treatment planning. In this work, we want to supplement the prediction model with a measurement of its uncertainty at each voxel. For this purpose, a full Bayesian approach would, however, be too costly. Instead, we compare, based on their correlation with the actual error, three simpler methods, namely, the dropout, the bootstrap and a modification of the U-Net. These methods can be easily adapted to other architectures. 200 patients with head and neck cancer were used in this work.

BackgroundIn cancer care, determining the most beneficial treatment technique is a key decision affecting the patient's survival and quality of life. Patient selection for proton therapy (PT) over conventional radiotherapy (XT) currently entails comparing manually generated treatment plans, which requires time and expertise.PurposeWe developed an automatic and fast tool, AI‐PROTIPP (Artificial Intelligence Predictive Radiation Oncology Treatment Indication to Photons/Protons), that assesses quantitatively the benefits of each therapeutic option. Our method uses deep learning (DL) models to directly predict the dose distributions for a given patient for both XT and PT. By using models that estimate the Normal Tissue Complication Probability (NTCP), namely the likelihood of side effects to occur for a specific patient, AI‐PROTIPP can propose a treatment selection quickly and automatically.MethodsA database of 60 patients presenting oropharyngeal cancer, obtained from the Cliniques Universitaires Saint Luc in Belgium, was used in this study. For every patient, a PT plan and an XT plan were generated. The dose distributions were used to train the two dose DL prediction models (one for each modality). The model is based on U‐Net architecture, a type of convolutional neural network currently considered as the state of the art for dose prediction models. A NTCP protocol used in the Dutch model‐based approach, including grades II and III xerostomia and grades II and III dysphagia, was later applied in order to perform automatic treatment selection for each patient. The networks were trained using a nested cross‐validation approach with 11‐folds. We set aside three patients in an outer set and each fold consists of 47 patients in training, five in validation and five for testing. This method allowed us to assess our method on 55 patients (five patients per test times the number of folds).ResultsThe treatment selection based on the DL‐predicted doses reached an accuracy of 87.4% for the threshold parameters set by the Health Council of the Netherlands. The selected treatment is directly linked with these threshold parameters as they express the minimal gain brought by the PT treatment for a patient to be indicated to PT. To validate the performance of AI‐PROTIPP in other conditions, we modulated these thresholds, and the accuracy was above 81% for all the considered cases. The difference in average cumulative NTCP per patient of predicted and clinical dose distributions is very similar (less than 1% difference).ConclusionsAI‐PROTIPP shows that using DL dose prediction in combination with NTCP models to select PT for patients is feasible and can help to save time by avoiding the generation of treatment plans only used for the comparison. Moreover, DL models are transferable, allowing, in the future, experience to be shared with centers that would not have PT planning expertise.