COVID-19 classification and identification method based on lung CT image

Abstract

The invention discloses a COVID-19 classification and identification method based on lung CT images, and the method comprises the steps: collecting a normal lung CT image, a lung tumor CT image and aCOVID lung CT image, obtaining three sample subsets, and forming a sample set; respectively pre-training three convolutional neural networks, namely AlexNet, GoogleNet and ResNet, by adopting a transfer learning method to respectively obtain initialization parameters of the three convolutional neural networks; respectively inputting the sample set into three pre-trained convolutional neural networks to obtain three individual classifiers; and integrating the three individual classifiers by adopting an ensemble learning method to obtain an ensemble classifier model. The overall classification performance of the integrated model is superior to that of an individual classifier, evaluation indexes such as specificity and sensitivity are high, and the requirement for rapid recognition of COVID-19 lung CT images can be well met.

Description

technical field [0001] The present invention relates to the technical field of medical image recognition, and more specifically relates to a COVID-19 classification and recognition method based on lung CT images. Background technique [0002] Due to the highly contagious nature of COVID-19, rapid and accurate detection of the pathogenic virus is of vital importance for subsequent treatment and isolation. Common detection methods for COVID-19 include nucleic acid reagent detection and CT examination. Clinical studies have shown that nucleic acid reagent detection has a high false negative rate for patients with suspected symptoms for the first time, and there are problems such as high requirements for the detection environment, strict procedures, and long time consumption, making it difficult to promote large-scale use. Computed tomography (CT) detection of COVID-19 has the characteristics of high sensitivity, low missed diagnosis rate, and high equipment popularity, which c...

AI Technical Summary

Problems solved by technology

[0003] At present, the COVID-19 virus is a serious threat to human life and health. With the sharp increase in the demand for medical treatment, the existing medical resources and diagnosis and treatment capabilities are not enough to deal with it. In addition, the density of personnel in hospitals in core epidemic areas has increased, and the risk of cross-infection has increased sharply.

Images