Ribs Suppression in Chest X-Ray Images by Using ICA Method

Abstract

In many developing countries such as Vietnam, Chest X-Ray (CXR) images are still one of the common tools to diagnose the lung cancer at the early phase. Early detection of lung nodules increases the chance of survival for a patient. In this study, we apply the Independent Components Analysis (ICA) to separate the ribs and other parts in lung images. First 20 files in Japanese Society of Radiological Technology (JSRT) database are used to perform the ICA algorithm, resulting in 90% of cases that the ribs are completely and partly suppressed, 85% of cases increases the nodule visibility. Two image enhancement methods are histogram equalization and Frangi filter are used as two artificially created inputs for ICA. This research also proposes the value of free parameters of Frangi filter.