Classification method and system of chronic sinusitis based on digital pathology slides

Abstract

The invention discloses a method for classifying chronic rhinosinusitis based on digital pathological glass slides, which comprises the following steps: image acquisition, acquiring digital pathological images from the nasal polyp slides of chronic rhinosinusitis and delineating them to generate a large mask image; Process to obtain a small pathological map and a small mask map; establish the training set data; establish a deep learning quantitative prediction module, use the Inception V3 model and train it on the ImageNet data set to obtain model parameters, and remove the last layer of the model. Fully connected layer FC, and add a fully connected layer FC with only one neuron in it, without any activation function, set the loss function to use the mean square error MSE, set the learning rate lr; Integrate the eosinophils of all small pathological pictures on the slide Proportional value to get the final auxiliary diagnosis result. The invention also discloses its system. The invention quickly obtains the proportion of eosinophils on pathological pictures through learning and training, and provides objective and high-accuracy auxiliary diagnosis results.

Description

technical field [0001] The invention relates to the technical field of medical treatment means, in particular to a method and system for classifying chronic sinusitis based on digital pathological slides. Background technique [0002] Chronic rhinosinusitis (CRS) can be clinically divided into two categories: without nasal polyps (CRSsNP) and with nasal polyps (CRSwNP). Chronic rhinosinusitis with nasal polyps (CRSwNP) is subdivided into two subtypes, eosinophilic nasal polyps (eCRSwNP) and non-eosinophilic nasal polyps (neCRSwNP). Eosinophilic nasal polyps (eCRSwNP) are sensitive to hormonal therapy, whereas non-eosinophilic nasal polyps (neCRSwNP) are sensitive to macrolide antibiotics. Clinically, for how to define eosinophilic nasal polyps (eCRSwNP) and non-eosinophilic nasal polyps (neCRSwNP), pathologists usually randomly take 10 high-power eosinophilic fields of nasal polyp slide specimens from patients. The average value of the proportion of granulocytes and the di...

AI Technical Summary

Problems solved by technology

In addition, due to different experiences of different doctors, or the same doctor at different times, the sampling estimates obtained by randomly selecting the field of view are not the same, that is, manual random sampling counts may also have measurement bias

[0004] Statistics on the proportion of eosinophils to inflammatory cells on the entire slide specimen can make a more accurate diagnosis and avoid sampling errors. However, it takes 2 to 4 hours for pathologists to complete the statistics of a slide specimen, and the time cost is extremely high

[0005] The clinical diagnosis of chronic sinusitis depends entirely on the experience of pathologists at this stage, lacking an objective, high accuracy and timeliness auxiliary diagnosis method and its system

Images