International Journal of Applied Mathematics and Machine LearningAuthor's: Linxuan Li, Zhendi Qin, Xuan Chen, Lie Chen, Fanzhi Kong and Fanbin Meng
Pages: [27] - [42]
Received Date: September 30, 2022
Submitted by: Jianqiang Gao
DOI: http://dx.doi.org/10.18642/ijamml_7100122264
Covid-19 epidemic have greatly increased the number of patients with lung disease, and physicians have difficulty assessing patients’ lung imaging with only personal experience and effort. To guarantee the efficiency of identification, it is necessary to establish a complete system for auxiliary lung disease identification. In response to the above problems, this paper will describe the process and results of a convolutional neural network (CNN) - based framework for lung disease image recognition. We randomly input image data using image data iterator and randomly selected a certain size of sample data for training in each batch. The system can identify chest X-ray images and lung CT images, and the identified lung diseases are Novel coronaviruses, Community-acquired pneumonia (CAP), and Viral pneumonia. The experimental test results of the classification system for image recognition in lung diseases have a high correct rate of 98.9%. From the experimental results, it is suggested that this system can assist physicians to complete the evaluation of lung imaging.
Covid-19, convolutional neural networks, deep learning, lung diseases image recognition classification system.
