The invention relates to a classification pre-judgment method and
system for a
brachytherapy source applicator
uterine cavity tube. The prediction method comprises the following steps: A, preprocessing a medical image; b, constructing a
deep learning classification prediction model; c, model training; d, outputting a result: inputting the independent test
verification set into a pre-judgment
system in sequence to obtain the probability of occurrence of each inclination angle; e, classification prediction result evaluation: based on a sklearn
database, respectively calculating an ROC curve lower area value AUCi of each inclination angle classification for an independent test
verification set and a model
training set; aUCi of all independent test
verification sets and model training sets is subjected to arithmetic averaging, and the result is used for evaluating the accuracy of
system prediction through a final AUC result. Through
artificial intelligence and
medical imaging technologies, before the source applicator is implanted, the optimal bending angle of the
uterine cavity tube is predicted and judged, the accuracy of the
prediction system is quantitatively evaluated, a decision basis is provided for implementation of clinical after-loading treatment, the pain of a patient is relieved, and the treatment efficiency is improved.