Department of computer engineering, Shabestar branch, Islamic Azad University, Shabestar, Iran
Online published on 20 March, 2014.
Alzheimer's disease is the most common form of neurodegenerative dementia worldwide. It affects brain structures even decades before manifesting its clinical symptoms. Whole brain atrophy rate is investigated in this study to reveal its discriminative power in differentiating between the Alzheimer's disease (AD) holders and normal controls (NC). In this way, a longitudinal study is performed on the Magnetic Resonance Images (MRI) of ADNI data set. A total of 60 subjects (30 AD and 30 controls) are participated in the study. Discriminative power of this measure is statistically analyzed and it is used as a feature in classifying subject using k-mean and FCM clustering algorithms. It is revealed that FCM has higher specificity besides k-mean but the same sensitivity. Accuracies of both systems are remarkable. It can be induced that pattern recognition algorithms even by unsupervised learning methods can help us to diagnose AD with considerable accuracy.
FCM, k-mean, Magnetic Resonance Image (MRI), Alzheimer's’ disease, Diagnostic, Longitudinal
