A classification system for cancer digital pathology cell images

Abstract

The present invention proposes a method and system for classifying cancer digital pathological cell images, performing block processing on the suspected lesion area of ​​interest, and performing feature extraction on the suspected lesion area after block processing by using local matching pattern texture features , and then use the extreme learning machine training method to classify and discriminate the extracted features to determine the benign and malignant tumors and grade division. The method and system for classifying cancer digital pathological cell images provided by the present invention utilizes local matching pattern texture features for feature extraction, analyzes cell texture features from both macro and micro aspects, and has rotation invariance, effectively overcoming the Many problems such as diversity and irregularity of shape provide reliable texture feature information for classification. At the same time, applying extreme learning machine to the classification of breast cancer cells shortens the training time, improves the speed of classification and discrimination, and improves the recognition accuracy.

Description

technical field [0001] The invention relates to the technical field of medical image processing, in particular to a method and system for classifying cancer digital pathological cell images. Background technique [0002] Breast cancer is a common malignancy among women in today's society. According to the latest statistics released by the World Health Organization (WHO) in February 2014 in the "World Cancer Report", in 2012, breast cancer was one of the three major cancers in the world (lung cancer, breast cancer, and colorectal cancer), accounting for the proportion of all cancers. 25%, ranking second. Therefore, how to prevent, diagnose and treat breast cancer more effectively, so as to reduce the damage of breast cancer to human beings, has become a very important topic in the medical field today. [0003] Immunohistochemical method has the advantages of high sensitivity and strong specificity, and has been widely used in basic research and clinical examination of patho...

AI Technical Summary

Problems solved by technology

At present, some computer-aided diagnosis methods have been used to improve the sensitivity and specificity of diagnosis. However, due to the differences in the preparation and staining methods of breast cell slices, the complexity of the background, the diversity and irregularity of cell shapes, etc., a large number of sample training is required, so it is impossible to obtain the classification results efficiently in terms of time. As a result, in terms of the accuracy of discrimination, the current method of feature extraction cannot make the classification results have a high accuracy.

Images