Machine Learning Approach for Tuberculosis Classification

The prevalence of tuberculosis, a severe sickness, is widespread in contemporary times. Given the severity of tuberculosis as a disease, it is imperative to adopt a proactive and expeditious approach to its treatment. Mycobacterium tuberculosis is a bacterial species responsible for the pathogenesis of tuberculosis. In order to overcome problem, a machinelearning methodology was proposed for diagnosis tuberculosis. Chest X-ray (CXR) images have served as a vital diagnostic tool for the respective medical condition. This study utilizes a threestep categorization procedure in order to ascertain the infection status of TB cases. The three steps of learning encompass histogram segmentation, GLCM feature extraction, and random forest classification. Histogram-based image segmentation is employed to partition the chest Xray image into distinct segments according to the pixel values. The GLCM technique has demonstrated efficacy in segmenting CXR pictures for illness diagnosis due to its ability to capture the grayscale connection between neighboring pixels through distance and angle measurements. The Modified Random Forest algorithm is a computational approach utilized to distinguish between tuberculosis-infected and normal CXR image data. The aforementioned approach provides the most precise methodology for diagnosing a patient's illness, with the random forest model achieving a 95.23% an accuracy rate.