EEG-Based EMG Estimation of Shoulder Joint for the Power Augmentation System of Upper Limbs

Abstract

Brain–Machine Interfaces (BMIs) have attracted much attention in recent decades, mainly for their applications involving severely disabled people. Recently, research has been directed at enhancing the ability of healthy people by connecting their brains to external devices. However, there are currently no successful research reports focused on robotic power augmentation using electroencephalography (EEG) signals for the shoulder joint. In this study, a method is proposed to estimate the shoulder’s electromyography (EMG) signals from EEG signals based on the concept of a virtual flexor–extensor muscle. In addition, the EMG signal of the deltoid muscle is used as the virtual EMG signal to establish the EMG estimation model and evaluate the experimental results. Thus, the shoulder’s power can be augmented by estimated virtual EMG signals for the people wearing an EMG-based power augmentation exoskeleton robot. The estimated EMG signal is expressed via a linear combination of the features of EEG signals extracted by Independent Component Analysis, Short-time Fourier Transform, and Principal Component Analysis. The proposed method was verified experimentally, and the average of the estimation correlation coefficient across different subjects was 0.78 (±0.037). These results demonstrate the feasibility and potential of using EEG signals to provide power augmentation through BMI technology.

Exploring Neurofeedback Training for BMI Power Augmentation of Upper Limbs: A Pilot Study

“…Furthermore, it is found that the training group that uses NFB training based on the alpha waves, beta waves, and theta waves performs better in the National Aeronautics and Space Administration (NASA) job tests and has a lower cognitive burden [ 33 ]. The results of our previous study [ 5 ] also showed that there are motion-related components in 0.1–40 Hz of EEG signals during the motion. Therefore, in this study, the frequency band (0–40 Hz) was used for NFB at first to find out the specific EEG changes of the elbow and the shoulder joints which have not been investigated yet, and then these changes will be used for NFB training and the verification of the training effect.…”

“…Until now, the EEG features and the methods for the BMI PA decoder haven’t been investigated extensively. We tried to extract the EEG features, and then successfully estimated electromyography (EMG) signals from EEG signals of the elbow joint [ 4 ], and the shoulder joint [ 5 , 6 ], respectively. With these estimated EMG signals, the BMI PA with an EMG-controlled PA device is expected to be realized.…”

Multimodal Imaging Genetic Analysis of Schizophrenia Based on Sparse Representation and Dictionary Learning

“…The biological significance of these latent variables in the field of imaging genetics is very ambiguous. The ICA can decompose the observed fMRI data into maximally independent signal sources and assume that the signal in the fMRI is a mixture of multiple independent components [12] . In a recent study, Calhoun et al [13] proposed group ICA and joint ICA.…”