AUTHOR=Lun Xiangmin , Yu Zhenglin , Chen Tao , Wang Fang , Hou Yimin 

TITLE=A Simplified CNN Classification Method for MI-EEG via the Electrode Pairs Signals

JOURNAL=Frontiers in Human Neuroscience

VOLUME=Volume 14 - 2020

YEAR=2020

URL=https://www.frontiersin.org/journals/human-neuroscience/articles/10.3389/fnhum.2020.00338

DOI=10.3389/fnhum.2020.00338

ISSN=1662-5161

ABSTRACT=The brain-computer interface (BCI), which is based on electroencephalography (EEG), provides independent information exchange and control channels for the brain and the outside world. However, EEG signals come from multiple electrodes, and the data of each electrode can extract multiple features. How to select electrodes and features to improve classification accuracy has become an urgent problem to be solved. This paper proposes a deep convolutional neural network (CNN) architecture with separated temporal and spatial filters, which selects the raw EEG signals of the electrode pairs over the motor cortex region as hybrid samples without any preprocessing or artificial feature extraction operations. It uses 5-layer CNN to learn EEG features, 4-layer max pooling to reduce dimensionality,  and a fully connected (FC) layer for classification. Dropout and batch normalization are used to solve the overfitting problem of the model. In the experiment, the 4 s EEG data of 10, 20, 60, and 100 subjects in the Physionet database are used as the data source, and the motor imaginations (MI) tasks are divided into four types: left fist, right fist, both fists and both feet. The results indicate that the global averaged accuracy on group-level classification can reach 97.28%, the area under the receiver operating characteristic (ROC) curve stands out at 0.997, and the electrode pair with the highest accuracy on 10 subjects dataset is FC3-FC4, with 98.61%. The research results also show that this work achieves high accuracy with a CNN classification approach using minimal (2) electrodes, which is the advantage compared to other methods on the same database. This proposed approach provides a new idea for simplifying the design of BCI system, and accelerates the process of clinical application.