Computed Tomography Image of Lung: A Visualization Enhancement Approach Based on Unsharp Marking and Thresholded Local Intensity Area Descriptor

Abstract

Viewing improvement is an important step in lung computed tomography (CT) image analysis. It assists to make diagnosis more accurate. In this paper, a new viewing improvement method for lung CT images is introduced. The proposed method consists of two stages. In the first stage, the image was deblurred and sharpened using the unsharp masking method. In the second stage, pre-processed image was enhanced the viewing by a modified method, thresholded local intensity area descriptor. The experiments were conducted on lung CT images from the LIDC-IDRI database. For the viewing improvement evaluation, our method was evaluated subjectively and quantitatively by five different measures of enhancement: Absolute mean brightness error, Edge content, Entropy, Peak signal-to-noise ratio, and Tenengrad criterion. The results clearly indicated that our approach was better than eight contrast enhancement methods. The proposed approach not only improved the contrast of the lung CT images, but also retained information essential for clinical diagnosis. Thus, it is expected to assist in improving the accuracy of diagnosis by physicians and computer-aided detection/diagnosis systems.