Lung segmentation using Support Vector Machine in computed tomography images

Abstract

Algorithms for extracting information about the structures present in an image are known as segmentation algorithms and play an important role in numerous biomedical applications where the images are the primary source of information. Image segmentation is a fundamental process in the area of medicine, although many alternatives have been proposed to solve the problem properly segment the objects that make up a scene, there is still one that can meet all the requirements that arise in this types of applications. Through this work we propose a method to segment, automatically, lungs obtained from computed tomography images through region growing. The starting point for the segmentation (seed) is obtained from the location of certain anatomical markers generated by a detection process implemented with Support Vector Machines least squares (LS-SVM) and interpolation of these by cubic spline in order to find a centroid of the contour obtained.