APPLICATION OF LUNG SEGMENTATION ON THORAX X-RAY IMAGES USING HOMOTOPY TREE METHOD

Abstract

The lungs are part of one of the organs in a system, namely breathing which is useful as a container for the exchange of oxygen from the air which replaces carbon dioxide in the blood. Which is in the chest and also the most important organs in the human body. And covered by a double-wall sac or pleura that lies on the outer surface of the lung and is also protected by the collarbone. Humans have two parts of the lungs, namely the right side is divided into 3 parts and the left is divided into 2 parts. Homotopy comes from the Greek homos and topos, tomos has the same activity while topos has topology. Homotopy Tree is a method used to obtain an object boundary in an image based on the color equation between neighbors. The test results with manual segmentation use the ROC (Receiver Operating Characteristic) method. In testing the system several X-Ray images of the throax for the right lung and left lung will produce an average value of accuracy, sensitivity and specificity of the two parts of the lungs. In this final project using several samples of x-ray images of the chest, which will be calculated the validation results of the right and left lung organs. This application system can automatically calculate the accuracy, sensitivity and specificity of human lungs which can simplify medical work such as assisting radiology. The purpose of the application system for lung segmentation is to facilitate the measurement of thorax x-ray images to be more efficient than manual measurement. This system uses the preprocessing stage to improve image quality in order to obtain good results.