A Histopathological Image Recognition Method

Abstract

The invention discloses a histopathological image recognition method, comprising the following steps: selecting non-disease and diseased training samples, non-disease and diseased test samples; combining the non-disease training samples and diseased training samples to establish a disease-free dictionary learning model and the diseased dictionary learning model, alternately iteratively optimize the two objective functions until the maximum number of iterations is reached, and learn the disease-free dictionary and the diseased dictionary; use the disease-free dictionary and the diseased dictionary to sparsely represent the test samples, and calculate The sparse reconstruction error vector of the test sample under the non-disease dictionary and the diseased dictionary; the classification statistics are obtained through the sparse reconstruction error vector, and the category of the test sample is determined by comparing the classification statistics with the threshold. The present invention proposes a new model and method for the application of dictionary learning in the classification of histopathological images. The learned dictionary with class labels has better sparse reconstruction and intra-class robustness for samples of the same class, and is more robust for samples of the same class. It has better inter-class discrimination.

Description

technical field [0001] The invention relates to a histopathological image recognition method. Background technique [0002] With the development of computer-aided diagnosis technology, the research of "digital pathology" has gradually attracted the attention of the majority of scientific researchers. Among them, how to accurately and automatically extract the discriminative features hidden in the image can provide a basis for subsequent histopathological image analysis or classification. The necessary information to quickly and accurately give disease grades and classifications has become one of the most challenging research topics in "digital pathology". [0003] Traditional feature extraction methods are mainly divided into the following two categories: the first category is based on specific domain or task-specific features, such as the size and shape of biological cells, grayscale or color information of images, texture, etc.; the second category Mainly focus on spatial...

AI Technical Summary

Problems solved by technology

[0006] However, due to the different characteristics presented by different types of histopathological images, the characteristics of cell morphology and geometric structure in the same type of histopathological images vary greatly, and the pathological features also show diversification, which leads to feature differences among samples of the same pathological image. It is larger than the feature difference between non-similar pathological image samples, so that the diseased dictionary learned by the above method is more similar to the disease-free dictionary, and the discrimination between the disease-free sample and the diseased sample is still low, and its classification performance is still to be determined. to improve

Images