Sung Chan Jun

Performance Variation in Motor Imagery Brain-Computer Interface: A Brief Review

Brain-computer interface (BCI) technology has attracted significant attention over recent decades, and has made remarkable progress. However, BCI still faces a critical hurdle, in that performance varies greatly across and even within subjects, an obstacle that degrades the reliability of BCI systems. Understanding the causes of these problems is important if we are to create more stable systems. In this short review, we report the most recent studies and findings on performance variation, especially in motor imagery-based BCI, which has found that low-performance groups have a less-developed brain network that is incapable of motor imagery. Further, psychological and physiological states influence performance variation within subjects. We propose a possible strategic approach to deal with this variation, which may contribute to improving the reliability of BCI. In addition, the limitations of current work and opportunities for future studies are discussed.

A review of brain-computer interface games and an opinion survey from researchers, developers and users.

In recent years, research on Brain-Computer Interface (BCI) technology for healthy users has attracted considerable interest, and BCI games are especially popular. This study reviews the current status of, and describes future directions, in the field of BCI games. To this end, we conducted a literature search and found that BCI control paradigms using electroencephalographic signals (motor imagery, P300, steady state visual evoked potential and passive approach reading mental state) have been the primary focus of research. We also conducted a survey of nearly three hundred participants that included researchers, game developers and users around the world. From this survey, we found that all three groups (researchers, developers and users) agreed on the significant influence and applicability of BCI and BCI games, and they all selected prostheses, rehabilitation and games as the most promising BCI applications. User and developer groups tended to give low priority to passive BCI and the whole head sensor array. Developers gave higher priorities to “the easiness of playing” and the “development platform” as important elements for BCI games and the market. Based on our assessment, we discuss the critical point at which BCI games will be able to progress from their current stage to widespread marketing to consumers. In conclusion, we propose three critical elements important for expansion of the BCI game market: standards, gameplay and appropriate integration.

Noise robustness analysis of sparse representation based classification method for non-stationary EEG signal classification

Abstract In the electroencephalogram (EEG)-based brain–computer interface (BCI) systems, classification is an important signal processing step to control external devices using brain activity. However, scalp-recorded EEG signals have inherent non-stationary characteristics; thus, the classification performance is deteriorated by changing the background activity of the EEG during the BCI experiment. Recently, the sparse representation based classification (SRC) method has shown a robust classification performance in many pattern recognition fields including BCI. In this study, we aim to analyze noise robustness of the SRC method to evaluate the capability of the SRC for non-stationary EEG signal classification. For this purpose, we generate noisy test signals by adding a noise source such as random Gaussian and scalp-recorded background noise into the original motor imagery based EEG signals. Using the noisy test signals and real online-experimental dataset, we compare the classification performance of the SRC and support vector machine (SVM). Furthermore, we analyze the unique classification mechanism of the SRC. We observed that the SRC method provided better classification accuracy and noise robustness compared with the SVM method. In addition, the SRC has an inherent adaptive classification mechanism that makes it suitable for time-varying EEG signal classification for online BCI systems.

Feasibility of approaches combining sensor and source features in brain–computer interface

Brain-computer interface (BCI) provides a new channel for communication between brain and computers through brain signals. Cost-effective EEG provides good temporal resolution, but its spatial resolution is poor and sensor information is blurred by inherent noise. To overcome these issues, spatial filtering and feature extraction techniques have been developed. Source imaging, transformation of sensor signals into the source space through source localizer, has gained attention as a new approach for BCI. It has been reported that the source imaging yields some improvement of BCI performance. However, there exists no thorough investigation on how source imaging information overlaps with, and is complementary to, sensor information. Information (visible information) from the source space may overlap as well as be exclusive to information from the sensor space is hypothesized. Therefore, we can extract more information from the sensor and source spaces if our hypothesis is true, thereby contributing to more accurate BCI systems. In this work, features from each space (sensor or source), and two strategies combining sensor and source features are assessed. The information distribution among the sensor, source, and combined spaces is discussed through a Venn diagram for 18 motor imagery datasets. Additional 5 motor imagery datasets from the BCI Competition III site were examined. The results showed that the addition of source information yielded about 3.8% classification improvement for 18 motor imagery datasets and showed an average accuracy of 75.56% for BCI Competition data. Our proposed approach is promising, and improved performance may be possible with better head model.

Achieving a hybrid brain?computer interface with tactile selective attention and motor imagery

OBJECTIVE We propose a new hybrid brain-computer interface (BCI) system that integrates two different EEG tasks: tactile selective attention (TSA) using a vibro-tactile stimulator on the left/right finger and motor imagery (MI) of left/right hand movement. Event-related desynchronization (ERD) from the MI task and steady-state somatosensory evoked potential (SSSEP) from the TSA task are retrieved and combined into two hybrid senses. APPROACH One hybrid approach is to measure two tasks simultaneously; the features of each task are combined for testing. Another hybrid approach is to measure two tasks consecutively (TSA first and MI next) using only MI features. For comparison with the hybrid approaches, the TSA and MI tasks are measured independently. MAIN RESULTS Using a total of 16 subject datasets, we analyzed the BCI classification performance for MI, TSA and two hybrid approaches in a comparative manner; we found that the consecutive hybrid approach outperformed the others, yielding about a 10% improvement in classification accuracy relative to MI alone. It is understood that TSA may play a crucial role as a prestimulus in that it helps to generate earlier ERD prior to MI and thus sustains ERD longer and to a stronger degree; this ERD may give more discriminative information than ERD in MI alone. SIGNIFICANCE Overall, our proposed consecutive hybrid approach is very promising for the development of advanced BCI systems.

Simple adaptive sparse representation based classification schemes for EEG based brain-computer interface applications

One of the main problems related to electroencephalogram (EEG) based brain-computer interface (BCI) systems is the non-stationarity of the underlying EEG signals. This results in the deterioration of the classification performance during experimental sessions. Therefore, adaptive classification techniques are required for EEG based BCI applications. In this paper, we propose simple adaptive sparse representation based classification (SRC) schemes. Supervised and unsupervised dictionary update techniques for new test data and a dictionary modification method by using the incoherence measure of the training data are investigated. The proposed methods are very simple and additional computation for the re-training of the classifier is not needed. The proposed adaptive SRC schemes are evaluated using two BCI experimental datasets. The proposed methods are assessed by comparing classification results with the conventional SRC and other adaptive classification methods. On the basis of the results, we find that the proposed adaptive schemes show relatively improved classification accuracy as compared to conventional methods without requiring additional computation.

Steady-State Somatosensory Evoked Potential for Brain-Computer Interface—Present and Future

Brain-computer interface (BCI) performance has achieved continued improvement over recent decades, and sensorimotor rhythm-based BCIs that use motor function have been popular subjects of investigation. However, it remains problematic to introduce them to the public market because of their low reliability. As an alternative resolution to this issue, visual-based BCIs that use P300 or steady-state visually evoked potentials (SSVEPs) seem promising; however, the inherent visual fatigue that occurs with these BCIs may be unavoidable. For these reasons, steady-state somatosensory evoked potential (SSSEP) BCIs, which are based on tactile selective attention, have gained increasing attention recently. These may reduce the fatigue induced by visual attention and overcome the low reliability of motor activity. In this literature survey, recent findings on SSSEP and its methodological uses in BCI are reviewed. Further, existing limitations of SSSEP BCI and potential future directions for the technique are discussed.

Calibration Time Reduction through Source Imaging in Brain Computer Interface (BCI)

Brain Computer Interface (BCI) is mainly divided into two phases; calibration phase for training and feedback phase. A calibration phase is usually time-consuming, thereby, being likely to raise subjects’ fatigue at the early stage. For more convenient and applicable BCI system it should be investigated to reduce such preparation (calibration) time before feedback phase. Beamformer is a source imaging technique widely used in MEG/EEG source localization problem. It passes only signals produced at the designated source point and filters out other signals such as noise. We conjecture information in source space may be consistent over well trained and good subjects. This idea facilitates to reuse existing datasets from the same or different subjects. Using IVa data in BCI competition III, we constructed a classifier from other 4 subject’s training data and performance was evaluated in source domain. In this work, we observed the proposed approach worked well, resulting in relatively good accuracies (73.21%, 74.21%) for two subjects.

Computational Study on Subdural Cortical Stimulation - The Influence of the Head Geometry, Anisotropic Conductivity, and Electrode Configuration

Subdural cortical stimulation (SuCS) is a method used to inject electrical current through electrodes beneath the dura mater, and is known to be useful in treating brain disorders. However, precisely how SuCS must be applied to yield the most effective results has rarely been investigated. For this purpose, we developed a three-dimensional computational model that represents an anatomically realistic brain model including an upper chest. With this computational model, we investigated the influence of stimulation amplitudes, electrode configurations (single or paddle-array), and white matter conductivities (isotropy or anisotropy). Further, the effects of stimulation were compared with two other computational models, including an anatomically realistic brain-only model and the simplified extruded slab model representing the precentral gyrus area. The results of voltage stimulation suggested that there was a synergistic effect with the paddle-array due to the use of multiple electrodes; however, a single electrode was more efficient with current stimulation. The conventional model (simplified extruded slab) far overestimated the effects of stimulation with both voltage and current by comparison to our proposed realistic upper body model. However, the realistic upper body and full brain-only models demonstrated similar stimulation effects. In our investigation of the influence of anisotropic conductivity, model with a fixed ratio (1∶10) anisotropic conductivity yielded deeper penetration depths and larger extents of stimulation than others. However, isotropic and anisotropic models with fixed ratios (1∶2, 1∶5) yielded similar stimulation effects. Lastly, whether the reference electrode was located on the right or left chest had no substantial effects on stimulation.

Computational study of subdural and epidural cortical stimulation of the motor cortex

Cortical stimulation (CS) has gained wide attention for its use in augmenting neurological recovery in various conditions. Noninvasive cortical stimulations using transcranial magnetic stimulation (TMS) and transcranial direct current stimulation (tDCS) are less invasive when delivering the electrical current to the patients brain, but have several limitations. Direct cortical stimulation (DCS) using an implantable stimulation system consisting of epidurally or subdurally placed electrodes and pulse generators, provides cortical stimulation and concurrent rehabilitative training in a stable fashion without limiting a patients activities. The effectiveness of these two types of DCS — epidural cortical stimulation (ECS) and subdural cortical stimulation (SCS) — has not been compared. In this work, a computer simulation study was conducted to predict the current density distributions (CDD) through cortical stimulations using subdurally or epidurally placed electrodes. The simulation study is based on the human motor cortex model with a three-dimensional finite element model (FEM). The change in CDD depending on the shape of the electrode (disc or ring) is discussed. The output current induced by SCS was about four times larger than that of ECS when voltage stimulations with the same magnitude were regulated. Thus, SCS showed substantially better penetration of the current into gray or white matter. Further, the ring electrode performed comparably or slightly inferior to the disc electrode in both cortical stimulations.