Title: Image classification of microscopic colonic images using textural properties and KSOM

Authors: Laurence A. Gan Lim, Raouf N.G. Naguib, Elmer P. Dadios, Jose Maria C. Avila

Addresses: Faculty of Engineering and Computing, Biomedical Computing and Engineering Technologies Applied Research Group (BIOCORE), Coventry University, Coventry, UK. ' Biomedical Computing and Engineering Technologies Applied Research Group (BIOCORE), Health Design and Technology Institute (HDTI), Coventry University, Coventry, UK. ' Manufacturing Engineering Department and Management, De La Salle University, Manila, Philippines. ' Department of Pathology, University of the Philippines, Manila, Philippines

Abstract: Colorectal cancer is considered the third most common neoplasm in the world. Traditionally, pathologists use a microscope to examine histopathological images of biopsy samples taken from patients and make judgments based on their professional expertise. Since this procedure is performed by a human expert, it is therefore subject to inconsistencies due to factors that might affect human performance. To overcome this problem, this paper proposes the use of Kohonen self-organising map and Haralick texture in the analysis of microscopic colonic images. The results presented here are preliminary and show great promise.

Keywords: KSOM; Kohonen self-organising maps; GLCMs; grey-level co-occurrence matrices; texture properties; colonic images; image classification; microscopic images; colorectal cancer; histopathological images; biopsy samples.

DOI: 10.1504/IJBET.2010.032698

International Journal of Biomedical Engineering and Technology, 2010 Vol.3 No.3/4, pp.308 - 318

Published online: 13 Apr 2010 *

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