Deep learning method for early Alzheimer disease diagnosis based on EEG signal

Alzheimer disease in its early stages, often known as a kind of dementia, is one of the major causes of death worldwide. It is a neurodegenerative illness in which brain electrical activity called electroencephalograph (EEG) slows down relative to that of healthy people. In the literature, a number of biomarkers are explored for detecting these EEG irregularities in signal. The EEG signals that record brain’s small electrical signal activity are highlighted in the current scientific study. EEG signals are utilized to diagnose Alzheimer disease based on relative power features. The wavelet transform is used for the denoising the EEG signal to increase the superiority of the signals. Multilayer perceptron and convolutional neural network were incorporated to learn the various features based on power for different stages of Alzheimer disease and normal people and achieved better results were compared to those found in the literature. The presented methodology gives the 96.72% accuracy on the real time database which is used in the study. As a result, the use of deep learning algorithms in clinical evaluation gives a baseline for investigation numerous neurological illnesses such as epilepsy, brain tumors, Alzheimer’s disease, and many other.