Abstract:Recent advances in artificial intelligence demand an automated framework for the development of versatile brain–computer interface (BCI) systems. In this article, we proposed a novel automated framework that reveals the importance of multidomain features with feature selection to increase the performance of a learning algorithm for motor imagery electroencephalogram task classification on the utility of signal decomposition methods. A framework is explored by investigating several combinations of signal decomposition methods with feature selection techniques. Thus, this article also provides a comprehensive comparison among the aforementioned modalities and validates them with several performance measures, robust ranking, and statistical analysis (Wilcoxon and Friedman) on public benchmark databases. Among all the combinations, the variational mode decomposition, multidomain features obtained with linear regression, and the cascade-forward neural network provide better classification accuracy results for both subject-dependent and independent BCI systems in comparison with other state-of-the-art methods. Impact Statement—The brain–computer interface (BCI) is a revolutionary device that utilizes cognitive function explicitly for the interaction of external devices without any motor intervention. BCI systems based on motor imagery have shown efficacy for stroke patient treatment, but poor performance, nonflexible characteristics, and lengthy training sessions have limited their use in clinical practice. The proposed automated framework overcomes these limitations. With the significant improvement of up to 26.1% and 26.4% in comparison with the available literature, the proposed automated framework could offer help to BCI device developers to develop flexible BCI devices and provide interaction for motor-disabled users.

Brain-Computer Interface Design Using Relative Wavelet Energy

A Brain-Computer Interface Based on a Few-Channel EEG-fNIRS Bimodal System

EEG Recognition Based on Multiple Types of Information by Using Wavelet Packet Transform and Neural Networks

A P300 Based Online Brain-Computer Interface System for Virtual Hand Control

Study on Classification Algorithms of a Brain—Computer Interface Based on Multichannel Linear Descriptors

BCI oriented EEG analysis using log energy entropy of wavelet packets

Research of P300 Feature Extraction Algorithm Based on ICA and Wavelet Transform

A CWT-based SSVEP classification method for brain-computer interface system

Linear Dynamical Systems Modeling for EEG-Based Motor Imagery Brain-Computer Interface.

Evaluation of Classical Features and Classifiers in Brain-Computer Interface Tasks

Practical Designs Of Brain-Computer Interfaces Based On The Modulation Of Eeg Rhythms

Pattern Recognition Of Motor Imagery Eeg Signal In Noninvasive Brain-Computer Interface

Motor Imagery Eeg Classification Using Wavelet Common Spatial Boosting Pattern

Common Spatial Pattern Method for Channel Selelction in Motor Imagery Based Brain-computer Interface.

Analysis Motion Imagination EEG Signal in Spatiotemporal-energy Domain.

Improved Discrete Wavelet Analysis and Principal Component Analysis for EEG Signal Processing

A P300-Based Bci Classification Algorithm Using Median Filtering and Bayesian Feature Extraction

Single-trial EEG Discrimination between Wrist and Finger Movement Imagery and Execution in a Sensorimotor BCI

Research on Feature Extraction Algorithm Commonly Used in Brain-computer Interface Technology

Toward the Development of Versatile Brain–Computer Interfaces

Efficacy and safety of adalimumab in Chinese adults with active ankylosing spondylitis: results of a randomised, controlled trial