State-Based Decoding of Continuous Hand Movements Using EEG Signals

Abstract

Recently, the advent of the non-invasive brain-computer interface (BCI) for continuous decoding of upper limb motions opens a new horizon for motor-disabled people. However, the performance of discrete-decoding BCIs based on discriminating different brain states are still more robust. In this study, we aimed to cascade a discrete state decoder with a continuous decoder to enhance the prediction of hand trajectories. EEG data were recorded from nine healthy subjects performing a center-out task with four orthogonal targets on the horizontal plane. The pre-movement data of each trial has been used for training a binary discrete decoder which identifies the axis of the movement based on common spatial pattern (CSP) features. Two non-parametric continuous decoders based on Gaussian process regression (GPR) have been designed for continuous decoding of hand movements along each axis using the envelope features of EEG signals in six frequency bands. In addition to those four principal orthogonal targets, some targets at random directions on the horizontal plane were recorded to evaluate the generalizability of the proposed model. The discrete decoder attained the average binary classification of 97.1% for discriminating movement along the x-axis and y-axis. The proposed state-based method achieved the mean correlation coefficient of 0.54 between actual and predicted trajectories for principal targets over all subjects. The trajectories of random targets were also decoded with a mean correlation of 0.37. The generalizability of the proposed paradigm proved by the findings of this study could open new possibilities in developing novel types of neuroprostheses for rehabilitation purposes.INDEX TERMS brain-computer interface, continuous decoding, electroencephalography, Gaussian process regression, state-based decoding.

Evaluation of temporal, spatial and spectral filtering in CSP-based methods for decoding pedaling-based motor tasks using EEG signals

“…In this context, some studies have reported classification of executed movements vs rest, where Kaiser et al reported an Acc close to 0.70 for classifying hand movements vs rest [39]. Hosseini et al used CSP based methods to design a rehabilitation device based on continuous 2D hand motion classification and decoding, where the classification results obtained an Acc close to 0.97 [15]. However, these two approaches have been used for the upper limbs.…”

“…An analysis was performed to verify the appropriate number of subjects for model generalization, based on the results reported in the literature. To address this, a G * power d = 2.13, a power of 0.90, and a value of α = 0.05 were used as recommended in previous studies focused on EEG and AM [15,19]. The minimum number of subjects computed by the G * Power software for the Wilcoxon signed-rank test was seven; however, a deviation of 20% was considered as the final sample size.…”

“…For instance, Luu et al developed a BCI system in which it was possible to decode walking motion and convert it into the control action of a walking avatar [14]. Hosseini et al used a strategy to detect current 2D hand motion to give visual feedback to the patient [15]. Amo et al focused on studying and comparing different frequency bands during execution and imagination of wrist motion [13].…”

“…This powerful technique has been widely used in the EEG field because, in addition to allowing high accuracy rates, it allows spatial filtering of channels, and thus, less processing time to train and evaluate classifiers. For instance, Hosseinni et al used CSP to classify handheld AMs to decode trajectories in a hybrid BCI to design more customized BCI controllers [15]. It is worth noting that this technique has been commonly used for upper limb classification tasks, but it has not been fully explored for lower limbs.…”

Influence of Temporal and Frequency Selective Patterns Combined with CSP Layers on Performance in Exoskeleton-Assisted Motor Imagery Tasks

“…Ten healthy individuals participated in this study (six male and four female), ranging in age from 21 to 48 years (29 ± 9), and without any neurological, psychological or other health conditions reported. The statistical power of the sample size was estimated using G*power software to suggest whether the results of the performance metrics can be compared with the literature [39,40]. Wilcoxon signed rank was used with a value of α = 0.05, an effect size of d = 2.13, and a sample size of n = 10.…”