Confidence selection-based cervical tissue pathology full-slide image automatic classification method

Abstract

The invention discloses a confidence selection-based cervical tissue pathology full-slide image automatic classification method, which comprises the steps of S1, segmenting a cervical tissue pathologyfull-slide image into small blocks with set sizes, collecting the small blocks generated by each full-slide image into a packet, and removing blank blocks in the packet; s2, building a CNN model; S3,training the CNN to specify the number of turns; S4, carrying out sequentially arranging and connecting to serve as a feature vector of the full-slide image; S5, training a support vector machine classifier; and S6, processing the cervical tissue pathology full-slide image needing to be classified through S1 and S4 to obtain a feature vector of the image, and inputting the feature vector into a trained support vector machine classifier to realize classification. According to the method, a confidence degree selection-based cervical tissue pathology full-slide image automatic classification method is used for assisting in providing reference for doctors, so that the misdiagnosis and missed diagnosis rates are reduced, the workload required by manual diagnosis is reduced, and the diagnosis efficiency is improved.

Description

technical field [0001] The invention relates to the field of automatic screening and analysis of cervical histopathological images, in particular to an automatic classification method for cervical histopathological whole slide images, which belongs to the application of machine learning neural network models in medical diagnosis. Background technique [0002] Cervical cancer is one of the most common gynecological malignancies and the fourth major factor that endangers women's physical and mental health. For the treatment of cervical cancer, early detection and early treatment are crucial. The process of cervical cancer development is normal tissue, intraepithelial precancerous lesions, and invasive carcinoma that breaks through the epithelial basement membrane. Among them, precancerous lesions refer to a series of lesions that occur before the occurrence of invasive cancer, also known as cervical intraepithelial neoplasia (CIN). To assess the severity of cervical precance...

AI Technical Summary

Problems solved by technology

[0006] The present invention aims at the problems of low efficiency and low accuracy of the existing cervical histopathological full-slide image automatic classification method in the above-mentioned background technology, improves on the basis of deep learning, and provides a method based on confidence selection An automatic classification method for cervical histopathological full-slide images, which is used to assist doctors in diagnosis, reduce their workload, improve diagnostic efficiency, and provide doctors with objective and accurate diagnostic results

Images