Radiotherapy three-dimensional dose prediction method based on deep learning

Abstract

The invention provides a radiotherapy three-dimensional dose prediction method based on deep learning, and the method comprises the steps: obtaining the CT data of a previous patient and a corresponding dose distribution diagram, so as to form a data set; preprocessing the data set, and dividing the data set into a training set, a verification set and a test set according to a set proportion to serve as a training sample, a verification sample and a test sample; establishing a U-Net neural network model, performing training learning by using the training sample, and performing cross validation on the U-Net neural network model according to the validation sample so as to adaptively adjust the structure of the U-Net neural network model; and inputting CT data of a patient to be predicted into the U-Net neural network model to obtain a predicted three-dimensional dose distribution diagram; finally, introducing Bayesian neural network structure to carry out model robustness testing, so that the clinical use safety and robustness of the model are improved. According to the invention, the quality and efficiency of tumor radiotherapy plan design can be improved, and the clinical treatment safety is improved.

Description

technical field [0001] The invention relates to the field of intelligent medical technology, in particular to a three-dimensional radiotherapy dose prediction method based on deep learning. Background technique [0002] The radiotherapy plan design for malignant tumors is one of the most critical steps in the radiotherapy process, and its quality determines the effect of the entire treatment. However, the traditional manual plan design is time-consuming and laborious, and the plan quality requires relatively high clinical experience of physicists. Different physicists It is prone to the problem of uneven quality of plan design among them. In order to ensure the efficacy and safety of patients and reduce tumor recurrence and normal tissue damage, dose prediction and evaluation of patient treatment plans are required. Therefore, how to realize the dose prediction and quality assessment of automatic radiotherapy planning, improve the quality and efficiency of tumor radiotherap...

AI Technical Summary

Problems solved by technology

[0003] The present invention provides a three-dimensional radiotherapy dose prediction method based on deep learning, which solves the problem that the existing radiotherapy dose to patients is prone to unevenness among different physicists, and can improve the quality and efficiency of tumor radiotherapy plan design. Increase the safety of clinical treatment

Images