چكيده به لاتين
Alzheimer's disease (AD) is the most common cause of dementia and major problem for the aged population. Mild cognitive impairment (MCI) is an early stage of Alzheimer's disease (AD). Early diagnosis of this disease increases the likelihood of treatment. In this study, we propose a new methods for detecting MCI by analyzing the Electroencephalogram (EEG) signal. The EEG signal is widely used because it is non-invasive, cheap, and convenient. The study was performed on two groups of 12 healthy subjects and 9 subjects with mild cognitive impairment. Two major approaches in feature extraction were used in this study. Band powers, common spatial pattern (CSP) and filter bank common spatial pattern (FBCSP) were employed as spectral, spatial and spatio-specteral filters for feature extraction in the first approach. In the second approach, different fuctional and effictive connectivity features were extracted from EEG signals. The extracted features in each approach were ranked by various criteria including mutual information, correlation and class center distance. Based on the results of this study, in the first approach, extracting features from spatio-spectral filters by FBCSP method outperformed other methods in classification accuracy. In the second approach, the results showed that using mutual information for measuring the connetivity between brain regions in all frequency sub-bands along with maximum class center distance criteria for features selection creates the best classification accuracy between two group healthy subjects and MCI patients.
Keywords: Mild Cognitive Impairment, Alzheimer's disease, Electroencephalogram, Common spatial pattern, Synchronization.
