![speedgoat kernel transfer matlab 2018b speedgoat kernel transfer matlab 2018b](https://www.mathworks.com/responsive_image/100/100/0/0/0/cache/matlabcentral/profiles/1642476_1522087715812_DEF.jpg)
Recently, Kwon and Im ( 2020) demonstrated that photobiomodulation before a BCI experiment could enhance the overall classification accuracy of fNIRS-based BCIs. The combination of fNIRS with other brain-imaging modalities also demonstrated a potential to improve the classification accuracy of the BCI system (Fazli et al., 2012 Shin et al., 2018b). ( 2020) proposed a general linear model-based preprocessing method to improve the classification accuracy of fNIRS-based BCI. For example, recent studies have reported significant improvements in the classification accuracy of fNIRS-based BCIs by employing high-density multi-distance fNIRS devices (Shin et al., 2017a) and using ensemble classifiers based on bootstrap aggregation Shin and Im ( 2020). Recently, many researchers have proposed new approaches to improve the performance of fNIRS-based BCIs. Previous studies (Coyle et al., 2007 Naseer and Hong, 2013 Hong et al., 2020) have reported that the performance of fNIRS-based BCI is high enough to be applied to practical binary communication systems that require a threshold classification accuracy of at least 70% (Vidaurre and Blankertz, 2010).
![speedgoat kernel transfer matlab 2018b speedgoat kernel transfer matlab 2018b](https://blogs.mathworks.com/images/simulink/2016Q2/mobile_3d_goat.gif)
During these mental tasks, increased cerebral blood flow caused by neural activities leads to an increase and decrease in ΔHbO and ΔHbR, respectively, which have been utilized to implement fNIRS-based BCIs (Ferrari and Quaresima, 2012 Schudlo and Chau, 2015). fNIRS can measure oxy- and deoxy-hemoglobin concentration changes (ΔHbO and ΔHbR) while an individual performs specific mental tasks such as mental arithmetic (MA), motor imagery (MI), mental singing, and imagining of object rotation. fNIRS is an optical brain-imaging technology used to record hemodynamic responses of the brain using near-infrared-range light of wavelength 600–1,000 nm. Recently, functional near-infrared spectroscopy (fNIRS), which is also one of the representative brain-imaging modalities, has attracted increasing attention owing to its advantages, including non-invasiveness, affordability, low susceptibility to noise, and portability (Naseer and Hong, 2015 Shin et al., 2017a).
![speedgoat kernel transfer matlab 2018b speedgoat kernel transfer matlab 2018b](https://www.speedgoat.com/Portals/0/adam/Content/kgjU80kLbkWIGogMvdn40Q/Image/mobile_single_default-1.jpg)
Various neuroimaging modalities such as electroencephalography (EEG), magnetoencephalography, and functional magnetic resonance imaging have been employed to implement BCIs (Mellinger et al., 2007 Sitaram et al., 2007 Hwang et al., 2013).
![speedgoat kernel transfer matlab 2018b speedgoat kernel transfer matlab 2018b](https://forces.embotech.com/Documentation/_images/speedgoat_high_level_3.png)
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.īrain–computer interfaces (BCIs) have been developed to decode a user's intention from their neural signals with the ultimate goal of providing non-muscular communication channels to those who experience difficulties communicating with the external environment (Wolpaw et al., 2002 Daly and Wolpaw, 2008). 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. 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%). Leave-one-subject-out cross-validation was employed to evaluate the average classification accuracy of the proposed 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. In this study, we propose a novel deep convolutional neural network (CNN)-based approach for implementing a subject-independent fNIRS-based BCI. 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. However, fNIRS signals are highly subject-specific and have low test-retest reliability. 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.