Performance Comparison of Deep Learning Models to Detect Covid-19 Based on X-Ray Images

Abstract

The SARS-Cov-2 outbreak caused by a coronavirus infection shocked dozens of countries. This disease has spread rapidly and become a new pandemic, a serious threat and even destroys various sectors of life. Along with technological developments, various deep learning models have been developed to classify between Covid-19 and Normal X-ray images of lungs, such as Inception V3, Inception V4 and MobileNet. These models have been separately reported to perform good classification on Covid-19. However, there is no comparison of their performance in classifying Covid-19 on the same data. This research aims to compare the performance of the three mentioned deep learning models in classifying Covid-19 based on X-ray images. The methods involve data collection, pre-processing, training, and testing using the three models. According to 2,169 dataset, the InceptionV3 model obtained an average accuracy value of 99.62%, precision value 99.65%, recall value 99.5%, specificity value 99.5%, and f-score value 99.52%; while the InceptionV4 model obtained an average accuracy value of 97.79%, precision value 98.11%, recall value 90.18%, specificity value 90.18%, and f-score value 97.25%; and the MobileNet model obtained an average accuracy value of 99.67%, precision value 99.77%, recall value 99.38%, specificity value 99.38%, and f-score value of 99.67%. The three models can classify the Covid-19 and Normal X-ray images based on research results, while the MobileNet model achieved the best performance. The model has stable performance in achieving graphic results and has extensive layers; the more layers there are to achieve better accuracy results.