Multiresolution wavelet analysis of ERPs for the detection of Alzheimer´s disease

Alzheimer´s disease, a neurological disorder usually seen in elderly people, is the most common of all cortical dementias. It can only be diagnosed with certainty via an autopsy. Neurologists usually diagnose Alzheimer´s disease from symptoms; however, misdiagnosis can be a problem. Additional techniques to increase the accuracy of ante-mortem diagnoses would be useful. In this study, event related potentials (ERPs) of two groups of patients were acquired. One group had been diagnosed as having Alzheimer´s disease and the other group as not having Alzheimer´s disease. The ERPs were analyzed using multiresolution wavelet analysis techniques. The analyzed signals were then used to train a multilayer perceptron (MLP) neural network to distinguish the signals belonging to the two groups of patients. Initial results demonstrated the feasibility of this approach