Cervical biopsy area recognition method and device based on multi-feature deep neural network

Abstract

The invention discloses a cervical biopsy region recognition method and device based on a multi-feature deep neural network. The device includes: an image acquisition unit for collecting physiological saline images, acetic acid images and iodine images of the cervix; a data processing unit including trained cervical biopsy regions The identification model, the cervical biopsy area identification model analyzes and processes the normal saline image, acetic acid image and iodine image, and outputs the probability label of the cervical biopsy area; the cervical biopsy area identification model includes: feature extraction layer, including 3 independent feature extraction sub- The network is used to extract the features of the saline image, acetic acid image and iodine image respectively; the feature combination layer is used to stitch together the three features; the top layer is used to identify the stitched features and output the probability label of the biopsy area of ​​the cervix; the display unit , get the said probability label and display it. The cervical biopsy area identification device can assist doctors to make accurate judgments on whether there is a biopsy area in the patient's cervix.

Description

technical field [0001] The invention relates to the field of medical image processing, in particular to a method and device for identifying cervical biopsy regions based on a multi-feature deep neural network. Background technique [0002] Cervical cancer is a common malignant tumor in gynecology, it is the second malignant tumor that seriously threatens women's health, and it is also the only malignant tumor with a clear etiology in humans. Colposcopy is a key link in cervical cancer screening and an accurate diagnosis of cervical lesions and cervical cancer. Early detection of cervical lesions can effectively reduce the risk of cervical cancer. [0003] The inspection steps of cervical lesions are mainly divided into three steps: (1) cervical cytology examination, the most commonly used is the Pap smear method; (2) colposcopy, if the cytology results are abnormal, colposcopy examination is required, observation Changes in cervical epithelial color, blood vessels, etc.; (3...

AI Technical Summary

Problems solved by technology

At present, many machine learning and image processing methods have been applied to the auxiliary field of colposcopy detection, including the detection of cervical os, the detection of vinegar white area, the prediction of cervical lesions, etc. These methods have played a certain auxiliary role, but cannot Fundamentally assist doctors to make more accurate judgments

And most of these methods only use 3%-5% acetic acid solution to colposcopic cervical images, which is different from doctors’ medical experience in judging whether there is a biopsy area by changing the image characteristics of physiological saline, 3%-5% acetic acid solution, and compound iodine solution. inconsistent

Images