Power frequency and wavelet characteristics in differentiating between normal and Alzheimer EEG

The diagnosis of Alzheimer's disease (AD), especially in its early stages, is becoming an increasingly important problem for clinical medicine as new therapies emerge. It seems likely that the progression of the disease can be significantly slowed with the use of medications early in the disease course. It will be also important to maintain current levels of sensitivity and specificity of the AD diagnosis as we move the diagnostic process earlier within the natural history of the disease. In the present study we compared power frequency and wavelet characteristics derived from electroencephalogram (EEG) in discriminating between AD patients and controls. We used these characteristics to train Learning Vector Quantization (LVQ) based neural networks to classify the AD/control subject groups. The results demonstrate the feasibility of this approach as a potential effective diagnostic tool for early Alzheimer's disease.