Covid-19 and Tuberculosis Detection in X-Ray of Lung Images with Deep Convolutional Neural Network

Abstract

Tuberculosis is an infectious disease with symptoms similar to those of Covid-19, such as fever, cough, and shortness of breath. Based on the existing cases, these two diseases attack the lungs and can affect their shape. Detection of this disease can be done through a chest X-ray. In the X-ray images of Covid-19 and Tuberculosis, both have ground-glass opacity and consolidation, thus classifying the two diseases is tricky if done manually. One method that can be used for classification is Convolutional Neural Network (CNN). The results obtained from this research are the implementation of the CNN algorithm with four convolutions which are convolution-pooling and repeated four times. The best architecture for parameters epoch 50 with the optimizer ADAM, image size 100x100 pixels, kernel size 3x3, and in the data scenario 80%:20%. The results of the level of accuracy of the classification process in the test data are 85.4%. In addition, the labeling prediction obtained is that the Covid-19 label is predicted to be correct with a probability percentage of 95.85%, while the probability percentage for the Tuberculosis label is 98%.