TUBERCULOSIS DETECTION METHOD AND LEVEL OF LUNG DAMAGE BASED ON CXR (CHEST X-RAY) USING INTELLIGENT COMPUTER VISION

Abstract

This invention relates to a method for detecting Tuberculosis (TB) and lung damage based on medical images of chest X-rays (CXR) using computer vision intelligence. This invention is intended to detect indications of TB while measuring and mapping the location of lung damage in patients. The method in this invention is able to automatically segment the lungs, detect areas of abnormalities typical of TB such as infiltrates, nodules, cavities, and pleural effusions, and provide location annotations on CXR images. In addition, this invention is also able to calculate the absolute area and percentage of lung involvement through the calculation of Absolute Lesion Proportion (ALP), so that the detection results are not only qualitative but also quantitative. Tuberculosis detection based on CXR images with output in the form of segmented images, lesion annotation results, and reports on the extent of lung damage obtained through computer vision intelligence, is covered by this invention.The computer vision intelligence used in this invention is a part of Artificial Intelligence.