Motor Imagery EEG Classification Using Capsule Networks

Various convolutional neural network (CNN)-based approaches have been recently proposed to improve the performance of motor imagery based-brain-computer interfaces (BCIs). However, the classification accuracy of CNNs is compromised when target data are distorted. Specifically for motor imagery electroencephalogram (EEG), the measured signals, even from the same person, are not consistent and can be significantly distorted. To overcome these limitations, we propose to apply a capsule network (CapsNet) for learning various properties of EEG signals, thereby achieving better and more robust performance than previous CNN methods. The proposed CapsNet-based framework classifies the two-class motor imagery, namely right-hand and left-hand movements. The motor imagery EEG signals are first transformed into 2D images using the short-time Fourier transform (STFT) algorithm and then used for training and testing the capsule network. The performance of the proposed framework was evaluated on the BCI competition IV 2b dataset. The proposed framework outperformed state-of-the-art CNN-based methods and various conventional machine learning approaches. The experimental results demonstrate the feasibility of the proposed approach for classification of motor imagery EEG signals.

Download Full-text

Motor Imagery Classification Using EEG Signals for Brain-Computer Interface Applications

Early Detection of Neurological Disorders Using Machine Learning Systems - Advances in Medical Technologies and Clinical Practice ◽

10.4018/978-1-5225-8567-1.ch013 ◽

2019 ◽

pp. 241-251

Author(s):

Subrota Mazumdar ◽

Rohit Chaudhary ◽

Suruchi Suruchi ◽

Suman Mohanty ◽

Divya Kumari ◽

...

Keyword(s):

Motor Imagery ◽

Nearest Neighbor ◽

Brain Computer Interface ◽

Computer Interface ◽

Test Cases ◽

Multiple Channels ◽

Eeg Signals ◽

Brain Computer Interfacing ◽

Electroencephalogram Eeg

In this chapter, a nearest neighbor (k-NN)-based method for efficient classification of motor imagery using EEG for brain-computer interfacing (BCI) applications has been proposed. Electroencephalogram (EEG) signals are obtained from multiple channels from brain. These EEG signals are taken as input features and given to the k-NN-based classifier to classify motor imagery. More specifically, the chapter gives an outline of the Berlin brain-computer interface that can be operated with minimal subject change. All the design and simulation works are carried out with MATLAB software. k-NN-based classifier is trained with data from continuous signals of EEG channels. After the network is trained, it is tested with various test cases. Performance of the network is checked in terms of percentage accuracy, which is found to be 99.25%. The result suggested that the proposed method is accurate for BCI applications.

Download Full-text

Electroencephalogram-Based Motor Imagery Classification Using Deep Residual Convolutional Networks

Frontiers in Neuroscience ◽

10.3389/fnins.2021.774857 ◽

2021 ◽

Vol 15 ◽

Author(s):

Jing-Shan Huang ◽

Wan-Shan Liu ◽

Bin Yao ◽

Zhan-Xiang Wang ◽

Si-Fang Chen ◽

...

Keyword(s):

Motor Imagery ◽

Recognition Accuracy ◽

Wavelet Packet ◽

Eeg Signals ◽

Convolutional Networks ◽

Learning Classifier ◽

Signal Features ◽

Electroencephalogram Eeg ◽

Bci System

The classification of electroencephalogram (EEG) signals is of significant importance in brain-computer interface (BCI) systems. Aiming to achieve intelligent classification of motor imagery EEG types with high accuracy, a classification methodology using the wavelet packet decomposition (WPD) and the proposed deep residual convolutional networks (DRes-CNN) is proposed. Firstly, EEG waveforms are segmented into sub-signals. Then the EEG signal features are obtained through the WPD algorithm, and some selected wavelet coefficients are retained and reconstructed into EEG signals in their respective frequency bands. Subsequently, the reconstructed EEG signals were utilized as input of the proposed deep residual convolutional networks to classify EEG signals. Finally, EEG types of motor imagination are classified by the DRes-CNN classifier intelligently. The datasets from BCI Competition were used to test the performance of the proposed deep learning classifier. Classification experiments show that the average recognition accuracy of this method reaches 98.76%. The proposed method can be further applied to the BCI system of motor imagination control.

Download Full-text

Reducing Response Time in Motor Imagery Using A Headband and Deep Learning

Sensors ◽

10.3390/s20236730 ◽

2020 ◽

Vol 20 (23) ◽

pp. 6730 ◽

Cited By ~ 1

Author(s):

Francisco M. Garcia-Moreno ◽

Maria Bermudez-Edo ◽

José Luis Garrido ◽

María José Rodríguez-Fórtiz

Keyword(s):

Deep Learning ◽

Response Time ◽

Motor Imagery ◽

Low Cost ◽

Interactive System ◽

Eeg Signals ◽

Hand Motion ◽

Left Hand ◽

Computer Interfaces ◽

Motion Imagery

Electroencephalography (EEG) signals to detect motor imagery have been used to help patients with low mobility. However, the regular brain computer interfaces (BCI) capturing the EEG signals usually require intrusive devices and cables linked to machines. Recently, some commercial low-intrusive BCI headbands have appeared, but with less electrodes than the regular BCIs. Some works have proved the ability of the headbands to detect basic motor imagery. However, all of these works have focused on the accuracy of the detection, using session sizes larger than 10 s, in order to improve the accuracy. These session sizes prevent actuators using the headbands to interact with the user within an adequate response time. In this work, we explore the reduction of time-response in a low-intrusive device with only 4 electrodes using deep learning to detect right/left hand motion imagery. The obtained model is able to lower the detection time while maintaining an acceptable accuracy in the detection. Our findings report an accuracy above 83.8% for response time of 2 s overcoming the related works with both low- and high-intrusive devices. Hence, our low-intrusive and low-cost solution could be used in an interactive system with a reduced response time of 2 s.

Download Full-text

An Evaluation of Different Fast Fourier Transform - Transfer Learning Pipelines for the Classification of Wink-based EEG Signals

Mekatronika ◽

10.15282/mekatronika.v2i1.5939 ◽

2020 ◽

Vol 2 (1) ◽

pp. 1-7

Author(s):

Jothi Letchumy Mahendra Kumar ◽

Mamunur Rashid ◽

Rabiu Muazu Musa ◽

Mohd Azraai Mohd Razman ◽

Norizam Sulaiman ◽

...

Keyword(s):

Fourier Transform ◽

Fast Fourier Transform ◽

Transfer Learning ◽

Test Accuracy ◽

Eeg Signals ◽

Computer Interfaces ◽

Physical Limitations ◽

Search Approach ◽

Electroencephalogram Eeg

Brain Computer-Interfaces (BCI) offers a means of controlling prostheses for neurological disorder patients, primarily owing to their inability to control such devices due to their inherent physical limitations. More often than not, the control of such devices exploits the use of Electroencephalogram (EEG) signals. Nonetheless, it is worth noting that the extraction of the features is often a laborious undertaking. The use of Transfer Learning (TL) has been demonstrated to be able to mitigate the issue. However, the employment of such a method towards BCI applications, particularly with regards to EEG signals are limited. The present study aims to assess the effectiveness of a number of DenseNet TL models, viz. DenseNet169, DenseNet121 and DenseNet201 in extracting features for the classification of wink-based EEG signals. The extracted features are then classified through an optimised Random Forest (RF) classifier. The raw EEG signals are transformed into a spectrogram image via Fast Fourier Transform (FFT) before it was fed into selected TL models. The dataset was split with a stratified ratio of 60:20:20 into train, test, and validation datasets, respectively. The hyperparameters of the RF model was optimised through the grid search approach that utilises the five-fold cross-validation technique. It was established from the study that amongst the DenseNet pipelines evaluated, the DenseNet169 performed the best with an overall validation and test accuracy of 89%. The findings of the present investigation could facilitate BCI applications, e.g., for a grasping exoskeleton.

Download Full-text