Using Deep Learning to Classify X-ray Images of Potential Tuberculosis Patients

Abstract

Deep Learning is widely used for image classification. Its success heavily relies on data which contains a sufficient amount of region of interest (~10%). However, due to the region of interest in medical images being as low as 1% of the entire image, Deep Learning has not been conveniently used for such cases. In this study, we employ recent techniques brought forth in Deep Learning and aim to classify X-ray images of potential Tuberculosis patients. Different types of learning rate enhancement techniques were used. Significant improvement was observed when coarse-to-fine knowledge transfer was employed to fine-tune the model further using multiple data augmentation techniques. We achieved an overall accuracy of 94.89% on the augmented images.