Title: Computer-aided abdomen disease diagnosis using curvelet-based texture analysis with progressive classification

Authors: Gaurav Sethi; Barjinder S. Saini

Addresses: Electronics and Communication Engineering Department, Dr. B. R. Ambedkar National Institute of Technology, Jalandhar 144011, Punjab, India ' Electronics and Communication Engineering Department, Dr. B. R. Ambedkar National Institute of Technology, Jalandhar 144011, Punjab, India

Abstract: Designing Computer-Aided Diagnostic (CAD) system for diagnosis diseases of various sizes is a difficult task. The abdomen diseases are of different sizes varying from medium to very small. Most CAD systems fail in classifying abdomen diseases of variable sizes. Hence, a method for classification of abdomen diseases of all sizes in computed tomography (CT) images is presented. The idea is to use features extracted using curvelet transform progressively for the classification of the abdomen diseases that leads to the accurate texture analysis and classification method. Experiments have been performed on 120 CT images of abdomen comprising of 30 images, each of normal tissues, cysts, tumours and stones. Three different sets of query images of variable sizes are used as input to the proposed system. Experimental result shows that the proposed method performs better than the conventional texture analysis and classification methods for diagnosis abdomen diseases of various sizes.

Keywords: abdomen diseases; CT images; computed tomography; computer-aided diagnostics; CAD; feature extraction; GLCM; progressive classification; image segmentation; wavelet transform; curvelet-based texture analysis; medical imaging; normal tissues; cysts; tumours; stones.

DOI: 10.1504/IJBET.2016.078335

International Journal of Biomedical Engineering and Technology, 2016 Vol.21 No.4, pp.322 - 342

Received: 26 Aug 2015
Accepted: 27 Nov 2015
Published online: 15 Aug 2016 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article