A Review on Tuberculosis Identification Using Convolutional Neural Networks

Tuberculosis is a deadly and pathogenic disease caused by bacterial Mycobacteria. It mostly affects our respiratory system. It can lead to death, if it is untreated. The World Health Organization has estimated that 0.0018 billions people were infected and died as a result of the disease. To solve this problem early diagnosis is very important. The proposed system introduces a methodology to find out tuberculosis by convolutional neural networks to help the doctors and patients. The network deals with inception v3 and uses dataset Montgomery and Shenzhen. Early diagnosis of TB, introduce a method which uses bayesian CNN. Efficient prediction of TB using chest X-Rays. The use of uncertainty in TB identification validates its results. On both datasets, the proposed method achieves 84.89% and 90.45% accuracy to minimize the error and achieve maximum forecast accuracy.