As clinical investigation on Alzheimer's disease (AD) based on large patient population has been increased, generalized methods to detect diagnostic features are required for physiological datasets recorded in different sites (i.e. different hospitals). This work aimed at using the multiway array decomposition (MAD) method to extract discriminative features from electroencephalograms (EEGs) in Alzheimer's disease (AD). We applied modified versions of three MAD methods (i.e. the parallel factor analysis (PARAFAC), Tucker3 model, and nonnegative tensor decomposition (NTD)) to multi-site recorded EEGs in AD and age- and sex-matched healthy subjects. Feed-forward multilayer Perceptron was used and trained to validate and optimize for classification of AD using two independent EEG databases. We showed, using another independent EEG dataset, that the MAD approach exhibited larger than 90% of classification accuracy for AD, which outperformed supervised spectral-spatial filters or other previous conventional EEG analyses.
|Number of pages||6|
|Publication status||Published - 2010|
|Event||2nd Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2010 - Biopolis, Singapore|
Duration: 14 Dec 2010 → 17 Dec 2010
|Conference||2nd Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2010|
|Period||14/12/10 → 17/12/10|