An Effective Method for the Detection of Tuberculosis Using Artificial Intelligence

Tuberculosis (TB), remains a significant global health challenge, is one of the deadliest diseases in the world, is affecting millions of people, especially in low and middle-income countries. The early detection of this disease is still a big problem to the affected due to the lack of technology used for detecting the disease at an earlier stage in the medical paradigm. One of the most frequently used screening tools is the chest X-ray (CXR) because it is non-invasive and prevents penetration to diagnose the disease. Even if this technique is non-invasive, the proper diagnosis is done with the help of expert radiologists; otherwise, human error is produced, and efficiency will be reduced. The main objective of this research work is to propose an Artificial Intelligence (AI) -based technique for the detection of TB through the automated analysis of chest X-rays. In this scenario, a new scheme using deep neural learning, especially Convolutional Neural Networks (CNNs), has been employed for high accuracy and specificity for classifying TB and normal subjects. The performance of the proposed system is analyzed with five parameters, mainly accuracy, precision, sensitivity, specificity and F1 score, used as diagnostic parameters for the disease diagnosis. This system can be used as a benchmark against the existing methods of TB detection and analysis. In addition to the above, this work help the subjects to improve their results in the diagnosis of TB using CXR and AI technology.