Medical Imaging for the Detection of Tuberculosis Using Chest Radio Graphs

Tuberculosis (TB) is the second largest death disease after Human Immunodeficiency Virus (HIV). In Pakistan as well as other countries radiologists faced a lot of problems in diagnosis the TB. The population of Pakistan is about 180,000,000 in which mostly peoples are poor. This paper presents our findings regarding this issue. In this work there is a complete description and review of the previous work done by the researchers. First Chest X-ray (CXR’s) images are segmented through random walker segmentation method and then a set of features on the base of intensity is computed. Computed features will help chest X-ray images to be classified on the bases of computed features as infected or healthy using SVM classifier. System performance is measured on dataset which is taken from Indiana University Hospital Network and data set collection is composed of 100 images. System accuracy is calculated through SVM classifier which is 73%. This research concludes with suggestions for further effort to more improve the quality of the system.