Author(s) :

Khushi Agarwal.

Volume/Issue :

Abstract :

Tuberculosis (TB) remains a significant global health challenge, leading to over 1.3 million deaths in 2022. Early and accurate detection of TB is crucial for effective treatment. Shortage of expert radiologists, especially in resource-constrained settings is one of the most significant impediments in detection of TB. This study explores the application of artificial intelligence in detecting TB features in chest X-rays. Utilizing a data set of 4,200 labeled chest X-ray images, a convolutional neural network (CNN) was developed to identify key radiological features associated with TB, including infiltrates, cavities, pleural effusion, enlarged lymph nodes, and miliary patterns. The model achieved an accuracy of 97.98%, demonstrating its potential as a supplementary tool for healthcare professionals. However, limitations such as reliance on a homogeneous dataset and lack of integration of patient medical history suggest further enhancements are necessary. Future work should focus on diversifying the dataset and incorporating comprehensive diagnostic elements to improve the model's applicability in varied clinical scenarios.