Early Detection of Tuberculosis Using SVM Machine Learning Algorithm

Abstract

Tuberculosis (TB) remains a significant global health concern, necessitating efficient diagnostic methods for early detection. This study proposes a comprehensive framework for the early detection of TB utilizing Chest X-Ray (CXR) images coupled with Computer-Aided Diagnosis (CAD) facilitated by Machine Learning (ML) techniques. The proposed framework integrates various stages including input image acquisition, preprocessing, edge detection, fuzzy C-means segmentation, feature extraction, and support vector machine (SVM) classification. Initially, CXR images are acquired and subjected to preprocessing to enhance their quality and remove noise. Subsequently, edge detection techniques are employed to highlight significant structures within the images. Fuzzy C-means segmentation is then applied to partition the lung region effectively, aiding in the isolation of potential TB-related abnormalities. Feature extraction is a crucial step wherein relevant attributes characterizing TB lesions are derived from segmented regions. These features encompass a diverse set of statistical, textural, and morphological descriptors, providing rich information for subsequent classification. Finally, an SVM classifier is trained on the extracted features to discriminate between TB-positive and TB-negative cases. The proposed framework demonstrates promising results in the early detection of TB from CXR images. Through the integration of ML algorithms, it offers automated and accurate diagnosis, potentially reducing the burden on healthcare professionals and facilitating timely interventions for TB patients. The effectiveness of the proposed methodology underscores its potential as a valuable tool in combating the spread of TB, particularly in resource-limited settings where access to expert radiologists may be limited.