Computer-aided diagnosis method for colorectal cancer based on small sample learning

Abstract

The invention discloses a computer-aided diagnosis method for colorectal cancer based on small sample learning. The computer-aided diagnosis method comprises the following steps: respectively extracting features of a pre-processed query image and a pre-processed support image set through two backbone networks with the same structure; performing cosine similarity calculation on the features, taking the obtained maximum similarity among all support pixels as a response value, performing normalization, converting the response value into an image size, and obtaining a similarity matrix corr as a response value feature map; and transmitting the features of the response value feature map, the query image and the support image to a subsequent convolutional network to obtain a segmentation and diagnosis result, calculating corresponding loss, and carrying out back propagation on an error to optimize and update network parameters. According to the method, the maximum similarity response value feature map is adopted, the guiding effect on segmentation is simpler and more effective, priori knowledge in the aspect of medical diagnosis is used for constraint, and the professionality and accuracy of the model in the aspect of diagnosis are improved.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a computer-aided diagnosis method for colorectal cancer based on small sample learning. Background technique [0002] Pathological slice images belong to the field of medical images, and the morphology of cell nuclei under the light microscope is still the main basis for tumor diagnosis. Cancer is a stubborn disease that is difficult for human beings to overcome, and colorectal cancer is a common malignant tumor in digestive tract diseases. Relevant data show that its incidence and fatality rate are among the top three cancers in my country, seriously affecting people's lives and health. Early detection of colorectal cancer is of great help to improve its cure rate. [0003] Most of the current detection methods rely on the personal experience of microscopes and pathologists, which makes the detection results subjective and variable. With the development of image proces...

AI Technical Summary

Problems solved by technology

Compared with natural images, medical images have relatively few data samples, and fine image annotation often requires a large number of experts to manually operate, which is time-consuming and labor-intensive. The high cost of manual annotation also limits the application of deep learning in the field of medical image segmentation. application

Images