Automated Electroencephalogram Temporal Lobe Signal Processing for Diagnosis of Alzheimer Disease

Nowadays early diagnosis of cognitive disease is a challenge because of lifestyle and stressful life. It is not easy to differentiate stress and Alzheimer condition from mere observation of personality. Detection of Alzheimer at the early stage is crucial to avoid later issues that include memory loss, person getting bedridden etc. The existing methods to detect Alzheimer is mostly expensive and requires laborious analysis which may result in delay for the start of actual medical treatment. Even some of the methods are invasive too. Hence there is demand to automate the detection of Alzheimer at the early stage using some noninvasive methods. In this chapter, the authors have suggested a strategy for the cognitive study of brain signals with Electroencephalogram (EEG) device. Collection of EEG signals are noninvasive and not much expensive. In our study, we have conducted experimentation for the temporal lobe. EEG signals originating from temporal lobes are down-sampled to 5 bands of varying frequency and further used as a dataset for the classification model. Five datasets were created from 5 different bands are used for training and testing purpose in proportion of 70–30 respectively. We have achieved praiseworthy accuracy for theta band for diagnosing of Alzheimer disease at early stage. We have achieved 98% accuracy by using deep learning model with 8 layers and activation function ReLu.