Cell grading system based on small sample recognition

Abstract

The invention relates to a cell grading system based on small sample recognition, and the system comprises a first obtaining module which is used for obtaining a training set, and the training set comprises known label cell samples; the training module is used for inputting the training set into the CNN model and performing training based on a Triplet Loss formula; the second acquisition module is used for acquiring an unknown tag cell sample; the first calculation module is used for calculating a similarity score between the unknown tag cell sample and the known tag cell sample by utilizing an embedded feature similarity function; the second calculation module is used for calculating the grade type score of the unknown label cell sample as the known label cell sample based on the similarity score; the building module is used for building a weighting model for the grade type score of the known label sample based on the unknown label sample, and building a graph structure through the weighting model; and the cell grading module is used for inputting the graph structure into a GCN model for cell grading. The method has the effect of improving the classification accuracy of the abnormal cells under the condition of a small number of training cell samples.

Description

technical field [0001] This application relates to the technical fields of artificial intelligence and medical treatment, and in particular to a cell grading system based on small sample identification. Background technique [0002] Cervical cancer is a malignant tumor that seriously endangers women's health and threatens women's lives. With the development of image processing and artificial intelligence technology, automatic pathological diagnosis technology has emerged. The key to this technology is to detect abnormal cells in the cervix in the images under the microscope and make a graded diagnosis of the abnormal cells in the cervix. [0003] The grading of abnormal cervical cells is carried out according to the TBS diagnostic criteria, and the abnormal cervical cells are further divided into categories such as ASC-US, ASC-H, LSIL, HSIL, SCC keratinization, and SCC non-keratinization. [0004] In related technologies, abnormal cytological screening based on image proces...

AI Technical Summary

Problems solved by technology

[0005] However, for small samples with a small number of abnormal cells and high similarity, due to the limited number of abnormal cells, the data of each type of sample is even more scarce when divided into several levels of abnormal cells

[0006] Furthermore, abnormal cells in small samples have many common features, and the similarity of various abnormal cells is high. When extracting features from abnormal cells, the most representative features cannot be effectively extracted, which reduces the accuracy of abnormal cell classification. Rate

Images