In this study, we focus on developing a computer - aided diagnosis (CAD) system which estimates dementia at home from electroencephalogram (EEG) data. In order to realize such a system, measurement with a small number of electrodes is necessary from the viewpoint of convenience. Furthermore, a short measurement time is required to reduce the burden on the patient. Therefore, we propose a method using average frequency and 3-electrode correlation value calculated from 1-minute data from 3 electrodes (P3, P4, Oz) on the occipital region at resting state with closed eyes. By using this method, it is shown that Alzheimer's disease and dementia with Lewy bodies (DLB) can be classified with an accuracy of 81% and 89% respectively.
