Title: Unsupervised segmentation of OSF by fusion of RGA and DCT with contextual information

Authors: Tathagata Ray, Anirban Mukherjee, J. Chatterjee, R.R. Paul, Pranab K. Dutta

Addresses: Department of Electrical Engineering, Indian Institute of Technology Kharagpur, Dist-West Midnapore, West Bengal 721302, India. ' Department of Electrical Engineering, Indian Institute of Technology Kharagpur, Dist-West Midnapore, West Bengal 721302, India. ' School of Medical Science and Technology, Indian Institute of Technology Kharagpur, Dist-West Midnapore, West Bengal 721302, India. ' Department of Oral and Maxillofacial Pathology, Gurunank Institute of Dental Science and Research, Panihati, Kolkata, 700114 West Bengal, India. ' Department of Electrical Engineering, Indian Institute of Technology Kharagpur, Dist-West Midnapore, West Bengal 721302, India

Abstract: The aim of this paper is to segment Light Microscopic (LM) images of Oral Sub-mucous Fibrosis (OSF) into its constituent layers. In this regard, fusion of features based on Region Growing Algorithm (RGA) and context-enhanced rotational invariant Discrete Cosine Transform (DCT) has been studied. The overall segmentation accuracy of this fused method is higher than that of context-enhanced DCT-based method. Fusion of features based on different methods often eliminates the disadvantages and utilises the advantages of individual method. Fuzzy c-means clustering has been found to be little ahead of k-means clustering in terms of segmentation accuracy.

Keywords: OSF images; oral submucous fibrosis; DCT; discrete cosine transform; region growing algorithm; RGA; fuzzy clustering; c-means clustering; feature fusion; segmentation accuracy; image segmentation.

DOI: 10.1504/IJBET.2010.034523

International Journal of Biomedical Engineering and Technology, 2010 Vol.4 No.2, pp.181 - 194

Published online: 07 Aug 2010 *

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