A DEEP LEARNING MODEL FOR PNEUMONIA DETECTION FROM X-RAY IMAGES

Abstract

The World Health Organization estimates that more than four million deaths oc- cur annually due to pneumonia and other diseases associated with air pollution, and the lat- est COVID-19 virus has dramatically increased the percentage of pneumonia cases. Over 150 million people get infected with pneumonia on an annual basis, especially children under 5 years old. There’s also a global shortage of radiologists in both developing and developed countries. Over 2/3 of people on earth do not have access to radiologists. According to the Association of American Medical Colleges, the U.S. is projected to have a shortage of 17,000 to 42,000 radiologists by 2033. Currently, the development of artificial intelligence and machine learning technologies, as well as the accumulation of large volumes of medical images, make it possible to create automated systems for analyzing medical images. The article presents a simple sequential model based on deep learning methods (convolutional neural networks) that helps detect pneumonia. X-ray images of the Women’s and Children’s Medical Center in Guangzhou were used for the model. The development of the pneumonia diagnostic program was carried out in Python. Training the neural network took 26 minutes and 12 epochs. The results obtained in the test data are: recall: 96%; precision: 92%; accuracy: 92%; and f1: 94% for pneumonia cases. This is no less than the result proposed in many popular works. The mod- el significantly reduces the burden on radiologists, helps them make decisions and save time, allows them to evaluate the quality of their work, and reduces the likelihood of medical errors.