https://www.scirp.org/journal/jbise J. Biomedical Science and Engineering, 2020, Vol. 13, (No. 7), pp: 153-167 https://doi.org/10.4236/jbise.2020.137015 153 J. Biomedical Science and Engineering Discrimination between Dementia Groups and Healthy Elderlies Using Scalp-Recorded-EEG-Based Brain Functional Connectivity Networks Sakura Nishijima 1 , Tetsushi Yada 1 , Toshimasa Yamazaki 1 , Yoshiyuki Kuroiwa 2 , Makoto Nakane 3 , Kimihiro Fujino 2 , Toshiaki Hirai 2 , Yasuhisa Baba 2 , Shoko Yamada 3 , Sho Tsukiyama 1 1 Department of Bioscience and Bioinformatics, Kyushu Institute of Technology, Fukuoka, Japan; 2 Department of Neurology, University Hospital, Mizonokuchi Teikyo University School of Medicine, Kanagawa, Japan; 3 Department of Neurosurgery, University Hospital, Mizonokuchi Teikyo University School of Medicine, Kanagawa, Japan Correspondence to: Toshimasa Yamazaki, Keywords: Electroencephalography, Network, Dementia, Synchronization Likelihood, Leave-One-Out Cross Validation, Euclidean Distance Received: June 11, 2020 Accepted: July 27, 2020 Published: July 30, 2020 Copyright © 2020 by author(s) and Scientific Research Publishing Inc. This work is licensed under the Creative Commons Attribution International License (CC BY 4.0). http://creativecommons.org/licenses/by/4.0/ ABSTRACT Objective: To establish a practical method for discriminating dementia groups and healthy elderlies, by using scalp-recorded electroencephalograms (EEGs). Methods: 16-ch EEGs were recorded during resting state for 39 dementia groups and 11 healthy elderlies. The connectivity between any two electrodes was estimated by synchronization likelihood (SL). The brain networks were constructed by normalized SL values. The present leave-one-out cross validation (LOOCV) required the Euclidean distance between any two subjects having 120-dimensional vectors concerned with the SL values for six frequency bands. In order to investigate factors which would affect the LOOCV results, principal component analysis (PCA) was applied to all the subjects. Results: The accuracy for the upper alpha yielded more than 80% and 70% in the dementia groups and the healthy elderlies, respectively. The LOOCV result could be explained in terms of brain networks such as executive control network (ECN) and default mode network (DMN) characterized by factor loadings of prin- cipal components. Conclusions: Dementia groups and healthy elderlies could be characte- rized by principal components of SL values between all the electrode pairs, even less con- nections, which revealed disruption and preservation of DMN and ECN. Significance: This study will provide a simple and practical method for discriminating dementia groups from healthy elderlies by scalp-recorded EEGs. Open Access