The invention relates to the technical field of
image processing, in particular to a P53
mutation classification method and
system for an
endometrial cancer pathological section, and the method comprises the steps: carrying out the
cutting of a full-view section of
endometrial tissue, obtaining a plurality of image blocks, and recording the initial position information of the image blocks; performing
feature extraction on each image block to obtain a
feature vector of each image block; inputting all the feature vectors and the initial position information into a feature classification model to obtain a corresponding
classification result; the feature classification model is used for performing sequence splicing on the feature vectors and inserting a category token to obtain a feature sequence; extracting context information of the feature sequence, and performing multi-scale
feature extraction on the context information to obtain a multi-scale comprehensive vector; and obtaining a corresponding
classification result based on the multi-scale comprehensive vector and all the initial position information. According to the embodiment of the invention, related personnel can be assisted to more accurately judge whether P53
mutation occurs in the
endometrial cancer pathological section or not.