Nodule detection from posterior and anterior chest radio graph using circular hough transform

Abstract

Lung cancer is the foremost cause of death in many regions of the world. Early detection betters the chances of survival. PA chest radiography is the most commonly used diagnosis tool for detecting lung tumor, because it is cost effective and requires less radiation dose. Radiologists fail to detect nodule from PA chest radio graphs, at early stage because of complex anatomical structure present in radio graphs. Computer aided diagnosis systems are developed to assist radiologist in detecting tumor from radio graphs at early stage. Paper describes the algorithms to find potential nodule from Posterior and Anterior (PA) chest radio graphic images. In this paper two algorithms were proposed to detect tumor from PA chest radio graphs. In the first method tumor is separated from radio graphic image using different techniques like threshold, region growing and morphological operations and identified using geometrical features extracted from the segmented tumor. In second method tumor detected automatically with threshold and Circular Hough transform.