Computer aided diagnosis of Alzheimer's disease from MRI brain images

Abstract

Computer aided diagnosis of Alzheimer's disease from MRI brain images has drawn the attention of pattern recognition and machine learning research community in last few years. Relevant feature extraction from such MRI images is one of the challenging issues in decision system. Recently dominant values obtained from 6 level decomposition using Slantlet transform are used to construct such features. In this paper, we have determined relevant features using first order statistics on coefficients obtained from Slantlet transform. We have compared the performance in terms of 8 well known and a combined performance measures. Experimental results on publicly available MRI dataset show that proposed method outperforms the dominant value based features extracted using Slantlet transform in terms of both individual and combined performance measures. © 2012 Springer-Verlag.