Spectro-Temporal EEG Modulation Energy based Diagnosis of Mild Alzheimer’s Disease

Alzheimer disease is the most common and fastest growing cognitive diseases in the United States. Developments of key biomarkers are challenging issues for diagnosis of Alzheimer disease and its progression. Previous research studies have highlighted the key use of EEG signals in diagnosis of several neurodegenerative disorders such as Alzheimer’s, Epilepsy and many more. In present paper, EEG spectro-temporal modulation energy approach is presented for diagnosis of Alzheimer’s disease. Firstly, the multi-channel EEG signal is separated into five bands: delta, theta, alpha, beta and gamma waves. Hilbert transformation is used for computing the temporal amplitude envelope of each band. The rate of modulation of each side band of each EEG signal is measured using modulation energy. These features are computed for 11 referential EEG signals as well as seven bipolar signals. For distinguishing the signal between two groups i.e. healthy control and Alzheimer subjected; Support vector machine classifier is used giving 95.71% accuracy on current database used.