Subject-Independent Functional Near-Infrared Spectroscopy-Based Brain–Computer Interfaces Based on Convolutional Neural Networks

Abstract

Functional near-infrared spectroscopy (fNIRS) has attracted increasing attention in the field of brain–computer interfaces (BCIs) owing to their advantages such as non-invasiveness, user safety, affordability, and portability. However, fNIRS signals are highly subject-specific and have low test-retest reliability. Therefore, individual calibration sessions need to be employed before each use of fNIRS-based BCI to achieve a sufficiently high performance for practical BCI applications. In this study, we propose a novel deep convolutional neural network (CNN)-based approach for implementing a subject-independent fNIRS-based BCI. A total of 18 participants performed the fNIRS-based BCI experiments, where the main goal of the experiments was to distinguish a mental arithmetic task from an idle state task. Leave-one-subject-out cross-validation was employed to evaluate the average classification accuracy of the proposed subject-independent fNIRS-based BCI. As a result, the average classification accuracy of the proposed method was reported to be 71.20 ± 8.74%, which was higher than the threshold accuracy for effective BCI communication (70%) as well as that obtained using conventional shrinkage linear discriminant analysis (65.74 ± 7.68%). To achieve a classification accuracy comparable to that of the proposed subject-independent fNIRS-based BCI, 24 training trials (of approximately 12 min) were necessary for the traditional subject-dependent fNIRS-based BCI. It is expected that our CNN-based approach would reduce the necessity of long-term individual calibration sessions, thereby enhancing the practicality of fNIRS-based BCIs significantly.

Dynamics of Long-Range Temporal Correlations in Broadband EEG During Different Motor Execution and Imagery Tasks

“…Since there were not enough trials of each condition per participant (22 trials) to train the classifier for each participant individually as above, we used the leave-one-participant-out cross-validation to train the LDA classifier. Leave-one-participant-out scheme gives participantindependent classifier performance on unseen data and is commonly used by several EEG and BCI studies such as Kwon and Im (2021) and Wu et al (2018). Thus, for each of the participants, a separate LDA classifier was trained for each sliding window as above with the feature vectors from corresponding windows from all the trials from the remaining 108 participants.…”

2020 International brain–computer interface competition: A review

“…Through the competition, we have confirmed that the presented issues were appropriate for assessing advances in BCIs; nevertheless, several technical concerns remain. For example, many scholars have solved few-shot learning, domain generalization, and cross-session problems with high levels of performance in other disciplines (Seo et al, 2020;Zhou K. et al, 2020;Kim G. et al, 2021;Kwon and Im, 2021;Li et al, 2021). BCI systems using intuitive speech imagination, also compared to those that require the imagination of existing behavior or visual external stimuli, have been found to lack sufficient decoding performance (Cooney et al, 2018;Lee et al, 2020a).…”

A novel EEG decoding method for a facial-expression-based BCI system using the combined convolutional neural network and genetic algorithm

“…Hence, selecting features relevant to the task will improve the classification performance. One advantage of CNN is the automatic extraction of discriminative features (Shajil et al, 2020 ; Kwon and Im, 2021 ). Learning hidden features and eliminating redundant information from the EEG signals will enhance the overall capability of BCI systems.…”

“…The pooling layer was inserted after the convolutional layer, to receive the compression feature map matrix from all selected channels and temporal values (Kwon and Im, 2021 ). The objective of the pooling layer is to improve the statistical efficiency of the network and improve its invariance (and subsequently its robustness).…”