Classification Method for Mild Cognitive Impairment Based on Power Variability of EEG Using Only a Few Electrodes

SUMMARY Patients with mild cognitive impairment (MCI) and healthy people were classified by using power variance function (PVF) of electroencephalography (EEG) recorded on only four electrodes in this study. PVF is an index of power variability of EEG proposed in a previous report. In this study, the balance of PVFs on a head of each person was analyzed. EEG signals of 25 MCI patients and those of 57 healthy people were recorded on 21 electrodes in accordance with the 10 to 20 International System while the patients were at rest with eyes closed for 5 min. After noise reduction, the balance of PVFs of the MCI patients and those of healthy people were compared. The results indicated that for MCI patients, the difference between PVFs of the frontal region and those of the occipital region was significantly smaller than that of healthy people in theta‐alpha band on four electrodes (O1, F3, Fz, and Cz as the reference). These electrodes were, therefore, selected for classification of MCI patients and healthy people. As the classification result, accuracy of 89.0%, sensitivity of 84.0%, and specificity of 91.2% were achieved.