Use of complexity based features in diagnosis of mild Alzheimer disease using EEG signals

Abstract

Alzheimer Disease is one of the neuro-degenerative diseases and most expensive disease in the modern society characterized by cognitive, intellectual as well as behavioral disturbance. Early diagnosis of Alzheimer disease is a major concern due to increasing population and lack of a standard and effective diagnosis. Although EEG is a powerful and relatively cheap tool for diagnosis of Alzheimer disease and dementia; it still not achieves the standards of clinical performance in terms of sensitivity and specificity to accept as a reliable technique for screening of Alzheimer disease. Hence, there is scope for increase in performance of EEG based diagnosis. This paper focuses on Electroencephalography based diagnosis and complexity based features for early diagnosis of Alzheimer disease. From the results, it is observed that the EEG of the Alzheimer patients slows down and it is less complex as that compared to the Normal patients. In this study, different complexity based features such as Spectral Entropy, Spectral Centroid, Spectral Roll-off and Zero Crossing Rate are used. K nearest Neighbor classifier is used for classifying the data between two groups i.e. Normal and Alzheimer disease patients giving accuracy of 96% in present research work.