컨볼루션 신경망 기반 운동심상을 이용한 뇌의 연결성 분석 및 분류방법 = CNN-based Classification of Brain Connectivity using Motor Imagery

국문 초록 (Abstract)

뇌-컴퓨터 인터페이스 (brain-computer interface; BCI)란 뇌에서 발생한 전기신호를 인공지능 알고리즘을 통해 사용자의의도를 예측하고, 그에 따라 로봇이나 컴퓨터를 제어해주는 기술로 세계 다양한 기관에서 미래 핵심 기술로 손꼽히는기술이다. BCI는 구현하는 방법 (Slow Cortical Potentials, Sensorimotor Rhythms, P300, Steady-State Visually Evoked Potential, Directional Tuning 등)에 따라 다양한 어플리케이션에 이용되고 있다. 하지만 BCI를 실생활에 사용하기 위해서는상황에 따라 시스템을 켜거나 꺼주거나 시스템의 모드 (typing, 로봇 제어, 전동 휠체어 제어 등)를 변경해주어야 한다. 본논문에서는 일반인 피험자 10명을 대상으로 EEG (Electroencephalography)를 측정 및 분석하여 피험자의 다양한 상태(resting, speech imagery, legs-motor imagery, hands-motor imagery)를 구분해내는 알고리즘을 개발하고 그 결과 88.25%의정확도로 상태를 구분할 수 있었다. 이는 BCI 모드 변경을 위한 핵심 알고리즘으로 BCI 기술의 실용화를 앞당길 것으로기대한다.

다국어 초록 (Multilingual Abstract)

The brain-computer interface (BCI) is a technology that predicts user’s intention through artificial intelligent algorithm and control robot or computer accordingly and is recognized as a core technology for the future by various organizations around the world. BCI is used in various applications according to the implementation method (Slow Cortical Potentials, Sensorimotor Rhythms, P300, Steady State Visually Evoked Potential, Directional Tuning, etc). However, to use BCI in real life, it is necessary to turn on/off the system according to the situation or to change the system mode (typing, robot control, electric wheelchair control, etc). In this paper, we developed an algorithm to measure various states (resting, speech imagery, legs-motor imagery, hands-motor imagery) of subjects by measuring and analyzing EEG in 10 subjects and as a result, we were able to distinguish the state with an accuracy of 88.25%. We expected that BCI technology would be put into practical use as a critical algorithm for changing BCI mode.

참고문헌 (Reference)

1 Dornhege G, "Toward Brain–Computer Interfacing" MIT Press 2007

2 S. B. Kotsiantis, "Supervised Machine Learning : A Review of Classification Techniques" 3-24, 2007

3 B. Burle, "Spatial and temporal resolutions of EEG : Is it really black and white? A scalp current density view" 97 : 210-220, 2015

4 C,J. Honey, "Network structures of cerebral cortex shapes functional connectivity on multiple time scales" 104 (104): 10240-10245, 2007

5 Velliste, M., "Motor cortical correlates of arm resting in the context of a reaching task and implications for prosthetic control" 34 (34): 6011-6022, 2014

6 Jirsa, Viktor K, "Handbook of Brain Connectivity" springer 2007

7 Yeom, H. G., "Estimation of the velocity and trajectory of three-dimensional reaching movements from non-invasive magnetoencephalography signals" 10 (10): 026006-, 2013

8 C, Chu, "Emergence of Stable Functional Networks in Long-Term Human Electroencephalography" 32 (32): 2703-2713, 2012

9 Cover, T.M, "Elements of Information Theory" Wiley edition 2006

10 M. Hutter, "Distribution of Mutual Information" 14 : 399-406, 2001

11 M. Rubinov, "Complex network measures of brain connectivity : Uses and interpretations" 52 : 1059-1069, 2010