A Comparative Study of Tuberculosis Detection Using Deep Convolutional Neural Network

Classification of distinct images is an engrossing research topic due to various purposes like disease detection and diagnosis, security purposes and so on. Image classification plays vital role in fast and early disease detection. Deep Learning is helping to a great extent in identifying images more efficiently. In this paper, four pre-trained models: DenseNet-169, MobileNet, Xception, Inception-V3 were used to compare the results of tuberculosis CXR detection. Two datasets: Shenzhen Hospital X-ray Set, Montgomery County X-ray Set were used and merged to achieve a better result. Data augmentation was used to make performance of the classifier more efficient. DenseNet169 performed best among the four architecture and got 91.6% validation accuracy, 92% precision, 92% recall, 92% F1-score and AUC score of 0.915. This research can have a positive impact to support doctors and researchers to identify tuberculosis and compare different model's performance.