Identification of Mycobacterium Tuberculosis Employing VGG-16 Feature Extraction and Classification Using Prominent Machine Learning Classifiers on X-rays

Tuberculosis, generally referred to as TB is a bacterial disease that affects the lungs and is one of the leading causes of death on a worldwide scale. TB is a bacterial illness that is caused by Mycobacterium tuberculosis. This lung disease is highly infectious and can potentially infect other human body organs. Early diagnosis and treatment of tuberculosis are essential components in lowering death rates and halting the progression of the illness. In this paper, we offer a system for diagnosing tuberculosis that takes X-ray images as its input, employs VGG16 as its feature extractor, and relies on machine learning (ML) classifiers to do its classification. Utilizing pre-trained convolutional neural networks (CNN), specifically VGG16, allowed for the successful extraction of features. The utilization of Adaboost yielded a notable accuracy of 95.11% and an F1 score of 0.921, representing the highest level of accuracy attained over the other ML classifiers. This study provides important insights into selecting ML algorithms for tuberculosis diagnosis. It is accomplished by using VGG16 feature extraction and classification with ML classifiers. If a sufficiently large dataset is provided, the proposed method has the potential to aid in tuberculosis detection.