Enhancing Tuberculosis Identification using Ensemble Deep Learning with Edge Detection

TB is caused by Mycobacterium tuberculosis. Early TB diagnosis saves deaths. Early TB detection sometimes involves chest x-rays. Lung cancer and TB seem same, making it hard for the radiologist to distinguish. This study describes an ensemble deep learning technique to TB diagnosis using Canny edge detected images and chest x-rays. This method adds variation to classifier errors by adding a new TB identifying characteristic. Features were first acquired from the original x-ray film, then from the edge detected picture. A deep learning system that incorporates data from several sources utilizing two deep learning-based methodologies is offered for automated TB detection. The integrated model architecture was validated using two public and one private datasets. The deep ensemble learning approach improved chest radiography prediction accuracy, even though both deep learning-based automated detection systems exceeded cutting-edge precision and specificity. X-ray images diagnosed TB with 99.67% accuracy, 99.2% precision, 98.99% sensitivity, 99.10% F1-score, and 99.5% specificity.