Inception V3 model for Tuberculosis detection using chest x-ray images
Shikha Prasher, Leema Nelson, S. Gomathi
Abstract
The respiratory disease tuberculosis is extremely deadly and morbid. Mycobacterium tuberculosis, which can infect many tissues, is the chronic disease that causes illness. Deep learning reveals value for a range of aspects of infectious diseases surveillance and detection, including that of the identification of tuberculosis. By use of a collected from various sources dataset on tuberculosis, a convolutional neural network model to evaluate the analysis and interpretation of the Inception-V3 deep learning model. By using data augmentation, batch normalization and deep learning classification algorithms, this research was able to accurately identify tuberculosis disease from chest X-ray photos. Transfer learning was utilized to train, assess, and evaluate InceptionV3 (CNN) model for the classification of tuberculosis and healthy cases. The Inception-V3 model which was accomplished got the best score, with an accuracy rate of 99%.