Enhancing Performance of a Hybrid EEG-fNIRS System Using Channel Selection and Early Temporal Features

Abstract

Brain-Computer Interface (BCI) techniques hold a great promise for neuroprosthetic applications. A desirable BCI system should be portable, minimally invasive, and feature high classification accuracy and efficiency. As two commonly used non-invasive brain imaging modalities, Electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) BCI system have often been incorporated in the development of hybrid BCI systems, largely due to their complimentary properties. In this study, we aimed to investigate whether the early temporal information extracted from singular EEG and fNIRS channels on each hemisphere can be used to enhance the accuracy and efficiency of a hybrid EEG-fNIRS BCI system. Eleven healthy volunteers were recruited and underwent simultaneous EEG-fNIRS recording during a motor execution task that included left and right hand movements. Singular EEG and fNIRS channels corresponding to the motor cortices of each hemisphere were selected using a general linear model. Early temporal information was extracted from the EEG channel (0-1 s) along with initial hemodynamic dip information from fNIRS (0-2 s) for classification using a support vector machine (SVM). Results demonstrated a lofty classification accuracy using a minimal number of channels and features derived from early temporal information. In conclusion, a hybrid EEG-fNIRS BCI system can achieve higher classification accuracy (91.02 ± 4.08%) and efficiency by integrating their complimentary properties, compared to using EEG (85.64 ± 7.4%) or fNIRS alone (85.55 ± 10.72%). Such a hybrid system can also achieve minimal response lag in application by focusing on rapidly-evolving brain dynamics.

Keywords: EEG; NIRS; general linear model; hybrid BCI; principal component analysis.

Figure 1

The experiment setup. (A) The environment of concurrent EEG-fNIRS measurement. The subject included in the figure was provided written, informed consent for the publication of this figure. (B) The paradigm used in the experiment. The “+” indicates the rest condition, the left arrow indicates left hand grasping task, and the right arrow indicates the right hand grasping task.

Figure 2

(A) Real photo of a subject wearing the cap completely mounted with EEG electrodes, fNIRS sources and detectors. (B) The configuration of the EEG electrodes and fNIRS optodes on the cap. Red circles denote the sources of fNIRS, green circles denote the detectors of fNIRS, the purple lines denote the fNIRS channels, and light blue and dark blue circles denote the EEG electrodes.

Figure 3

(A) Group-wise location summary of the selected EEG and fNIRS channels for all subjects. (B) Zoom-in view of the group-wise summarized location. An orange triangle represents a pair of selected fNIRS channel and their corresponding EEG channel. The number in the triangle represents the number of subjects whose selected channel is located at the given area.

Figure 4

The flow chart of the study.

Figure 5

Classification accuracies of two hand movements obtained from three feature sets (EEG + fNIRS, EEG-only, and fNIRS-only).

Figure 6

Statistical plot of the classification accuracies obtained from the three feature sets, respectively. The asterisk “^*” indicates a significant difference (p < 0.5, t_{10, 0.975} denotes the critical value with 10 degree of freedom and significant level = 0.05).

Figure 7

Example (Subject 2) of the average HbO and HbR signals of a selected channel on left hemisphere before (A,C) and after (B,D) the PCA denoising. The “0” denotes the onset of the stimuli. LH: Left Hand (blue); RH: Right Hand (red).