Tuberculosis or TB is an infectious disease caused by the bacteria Mycobacterium tubercolusis. This disease usually attacks the lungs, but can also affect other organs such as the kidneys, bones and brain. TB is highly contagious, and can spread through the air when someone who is infected coughs or sneezes. Risk factors that can increase a person's chances of developing TB include a weak immune system, such as people with AIDS, diabetes, or people taking immunosuppressant drugs. And people who live or work in environments with high rates of TB transmission are also at risk of infection. Symptoms of TB are usually a cough that lasts more than three weeks, unexplained weight loss, fever, night sweats and persistent fatigue. In more severe cases, TB can cause coughing up blood, chest pain and difficulty breathing. One of the examination tools that can be used to detect TB disease is x-rays. Which produces X-Rays to help and confirm the diagnosis of TB disease, to see the chest part of the body which is used as medical record documentation. In X-ray photos, random dark and light spots of noise are often found which are caused by several factors. Based on the facts above, image segmentation is an important task for doctors in diagnosing disease. Automatic detection or segmentation of lung images from chest x-ray images is the initial stage of the diagnosis process. This research aims to implement a segmentation method to determine edge detection in clearer images using several segmentation methods, namely the Canny Edge Detection method, Sobel reading chest x-ray results for tuberculosis. And canny edge detection with segmented RGB image (otsu's thresholding) produces the highest value, namely 230,466.0 pixels and a lesion volume of 14,818.625 mm3.
Read full abstract