Title: A combination of dual-tree discrete wavelet transform and minimum redundancy maximum relevance method for diagnosis of Alzheimer's disease

Authors: Namita Aggarwal; Bharti Rana; R.K Agrawal; Senthil Kumaran

Addresses: School of Computer & Systems Sciences, Jawaharlal Nehru University, New Delhi 110067, India ' School of Computer & Systems Sciences, Jawaharlal Nehru University, New Delhi 110067, India ' School of Computer & Systems Sciences, Jawaharlal Nehru University, New Delhi 110067, India ' Department of NMR, All India Institute of Medical Sciences, New Delhi 110029, India

Abstract: In this paper, we propose a three-phased method for diagnosis of Alzheimer's disease using the structural magnetic resonance imaging (MRI). In first phase, gray matter tissue probability map is obtained from every brain MRI volume. Further, five regions of interest (ROIs) are extracted as per prior knowledge. In second phase, features are extracted from each ROI using 3D dual-tree discrete wavelet transform. In third phase, relevant features are selected using minimum redundancy maximum relevance features selection technique. The decision model is built with features so obtained, using a classifier. To evaluate the effectiveness of the proposed method, experiments are performed with four well-known classifiers on four data sets, built from a publicly available OASIS database. The performance is evaluated in terms of sensitivity, specificity and classification accuracy. It was observed that the proposed method outperforms existing methods in terms of all three performance measures. This is further validated with statistical tests.

Keywords: Alzheimer's disease; diagnosis; classification; dual-tree DWT; discrete wavelet transform; brain scanning; MRI; magnetic resonance imaging; mRMR; minimum redundancy maximum relevance; bioinformatics.

DOI: 10.1504/IJBRA.2015.071944

International Journal of Bioinformatics Research and Applications, 2015 Vol.11 No.5, pp.433 - 461

Received: 31 Jan 2015
Accepted: 14 Jun 2015
Published online: 24 Sep 2015 *

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