Diagnosis of tuberculosis based on deep learning

Tuberculosis is one of the most dangerous diseases in the world. It is caused by the bacterium Mycobacterium bacteria. Pulmonary tuberculosis is the most common form of TB. The bacteria primarily target the lungs, resulting in symptoms such as an enduring cough (occasionally accompanied by blood-tinged mucus) and a reduction in body weight. If not addressed, it can lead to extensive harm to the lung tissue. Accurate diagnosis is important to control the disease. To achieve this goal a computer-aided detection system is created. Chest X-ray Images are used to train the proposed model. We have applied multiple machine learning algorithms (Logistic Regression, KNN, SVM, Ensemble Learning). CNN gave the best accuracy 99.5%. The kernel size of 3×3 is used for the convolution layer and 2×2 is used for the pooling layer. Adam optimizer helped to get the best result.