Method and system for assisting in building radiomics model based on semi-supervised learning

The invention discloses a method for assisting in building a radiomics model based on semi-supervised learning. The method comprises the following steps: acquiring a to-be-trained data set with a label and a to-be-trained data set without a label; collecting a data set without a label from the internet according to the format and the classification purpose of the data with the label; preprocessing a data set with a label and a data set without a label; randomly dividing the labeled data set into a training set and a test set; fitting a plurality of classifiers for the training set, performing voting and scoring, adjusting a label threshold value, and labeling non-label data; screening the features of the training set, and screening the features of the training set and the whole labeled data; and classifying and fitting the features in the training set, and classifying and de-fitting the features in the training set and the labeled data set. According to the method, the stability of the repeated experiment eff