An artificial intelligence-assisted reading system for cervical cell liquid-based smear

Abstract

The invention discloses an artificial intelligence-assisted reading system for cervical cell liquid-based smears. The system includes: a cell image acquisition module, a cell image preprocessing module, and a cell image detection and segmentation module, which are used for different cell components of image cells Carry out automatic detection, and automatically segment the nucleus, cytoplasm, and background in the same cell component at the same time; fine-tune and optimize the segmentation results; the cell rapid classification recognition module is used to identify the segmented image and distinguish it into single cells or cell clusters Clusters; use the dual-stream convolutional neural network in the additional knowledge field and the constructed cell knowledge map to classify and identify individual cells, and obtain two classification results respectively; use the dual-stream convolutional neural network model of cell clusters to realize the identification of inseparable cell clusters identification; and the interpretation and post-processing module, which are used for joint interpretation of the two classification results of a single cell, and conflict handling, to obtain the classification result of a single cell.

Description

technical field [0001] The invention relates to the field of medical cell image processing, in particular to an artificial intelligence-assisted reading system for cervical cell liquid-based smears. Background technique [0002] Cervical cancer is a malignant tumor that occurs in the cervix, and it is closely related to high-risk HPV infection. Early detection, diagnosis and treatment of cervical precancerous lesions are one of the strategies to reduce the incidence and mortality of cervical cancer. Cervical cancer screening is simple and economical. Many countries use screening methods to control the incidence and mortality of cervical cancer and reduce the burden of disease. [0003] From 2008 to 2016, more than 2 million free cervical cancer screenings were completed in Beijing, more than 4,000 cases of cervical cancer and precancerous lesions were detected, and the positive detection rate of cervical cytology was 2.3%. According to foreign data and some regions in my co...

AI Technical Summary

Problems solved by technology

[0003] From 2008 to 2016, more than 2 million free cervical cancer screenings were completed in Beijing, more than 4,000 cases of cervical cancer and precancerous lesions were detected, and the positive detection rate of cervical cytology was 2.3%. According to foreign data and some regions in my country According to the research data, experts speculate that the positive detection rate of cervical cytology in the normal population in my country is roughly between 5-7%, but the current detection rate in Beijing is far from this standard

[0006] The existing quantitative analysis and auxiliary diagnosis methods of cervical cells still have obvious defects, and the research on cervical cell analysis technology is still in its infancy. Neighborhood information and color distribution information establish accurate cell image segmentation, make full use of the cervical cell database and cervical cell clinical diagnosis rules to establish a framework for quantitative analysis of cervical cells and intelligent auxiliary diagnosis, and provide an artificial intelligence-assisted film reading system based on liquid-based cervical cell smears Provide software platform

Images