Texture Characterization of CT Images Based on Ridgelet Transform

Human vision system has limitations in distinguishing the broad range of gray level values. Human eye can discriminate pixel intensities up to 15-30 gray levels. This restricts the qualitative analysis of radiological images. Hence quantitative analysis is preferred to reveal more information from the image. This article presents texture feature based approach for biomedical image analysis. Images acquired from Computerized Tomography (CT) scan machine have been used for the work. A novel method of texture feature extraction based on Ridgelet transform has been reported in this paper. In the first step, work involves determination of texture features from Region of Interest (ROI). Energy and entropy in partitions of Ridgelet transform images represent texture features. During the next step of work, two-class and multi-class classification has been carried out. Percentage Correct Classification for Ridgelet based energy and entropy features and comparative analysis of performance measures for different organ images have been reported.