A typical BCI system consists of five stages: brain-signal acquisition, preprocessing, feature extraction/selection, classification, and application interface. In the first brain-signal acquisition stage, suitable signals are acquired using an appropriate brain-imaging modality. Since the acquired signals are normally weak and contain noises (physiological and instrumental) and artifacts, preprocessing is needed, which is the second stage. In the third stage, some useful data so called “features” are extracted. These features, in the fourth stage, are classified using a suitable classifier. Finally, in the fifth stage, the classified signals are transmitted to a computer or other external devices for generating the desired control commands to the devices. In neurofeedback applications, a real-time display of brain activity is desirable, which enables self-regulation of brain functions. The figure below depicts a schematic of (hybrid) functional near-infrared spectroscopy (fNIRS) and electroencephalography (EEG) BCI.

[Schematic of a hybrid fNIRS-EEG BCI]

For more details, see “fNIRS-based brain-computer interfaces: a review” Frontiers in Human Neuroscience , Vol. 9, AN: 3, pp. 1-15, January 2015.

In this study, we present a systematic method to reduce the time lag in detecting initial dips using a vector-based phase diagram and an auto-regressive moving average with exogenous signals (ARMAX) model-based q-step-ahead prediction algorithm. With functional near-infrared spectroscopy (fNIRS), signals related to mental arithmetic and right-hand clenching are acquired from the prefrontal and left primary motor cortices, respectively. The interrelationship between oxygenated hemoglobin, deoxygenated hemoglobin, total hemoglobin and cerebral oxygen exchange are related to initial dips.

[Application of initial dips for BCI]

[Time series (ΔHbO, ΔHbR, ΔHbT, ΔCOE) for 20 s after the stimulus for all 5 trials of the mental arithmetic task]

Specifically, a threshold value from the resting state hemodynamics is incorporated, as a decision criterion, into the vector-based phase diagram to determine the occurrence of initial dips. To further reduce the time lag, a q-step-ahead prediction method is applied to predict the occurrence of the dips. A combination of the threshold criterion and the prediction method resulted in the delay time of about 0.9 s. The results demonstrate that rapid detection of initial dip is possible and therefore can be used for real-time brain-computer interfacing.

[Vector-based phase diagram of all 12 channels for the mental arithmetic task]

[Magnitudes of vectors from measured and predicted data for the mental arithmetic task]

For more details, see “Reduction of delay in detecting initial dips from functional near-infrared spectroscopy signals using vector-based phase analysis,” International Journal of Neural Systems, 2016. ( DOI: 10.1142/S012906571650012X)

Deception is a human behavior that many people experience in daily life. It involves complex neuronal activities in addition to several physiological changes in the body. A polygraph, which can measure some of the physiological responses from the body, has been widely employed in lie-detection. Many researchers, however, believe that lie detection can become more precise if the neuronal changes that occur in the process of deception can be isolated and measured. In this study, we have combined both measures (i.e., physiological and neuronal changes) for enhanced lie-detection. Specifically, to investigate the deception-related hemodynamic response, fNIRS is applied at the prefrontal cortex besides a commercially available polygraph system. A mock crime scenario with a single-trial stimulus is set up as a deception protocol.

[Block diagram of measurement setup]

The acquired data are classified into “true” and “lie” classes based on the fNIRS-based hemoglobin-concentration changes and polygraph-based physiological signal changes. Linear discriminant analysis is utilized as a classifier. The results indicate that the combined fNIRS-polygraph system delivers much higher classification accuracy than that of a singular system.

[Comparison of classification accuracies of individual modalities and the combined system]

The body of BCI research can be categorized into active, reactive and passive brain tasks. In a passive BCI, an arbitrary activity generated without any objective control is used: Drowsiness (a passive brain activity caused by fatigue or sleep deprivation) is a significant contributor to traffic accidents. In this lab, we use functional near-infrared spectroscopy (fNIRS) to discriminate the alert and drowsy states for a passive BCI.

[The experimental scheme used for drowsiness detection]

The passive brain signals for the drowsy state are acquired from the prefrontal and dorsolateral prefrontal cortex. The experiment is performed using a driving simulator, and their brain activity is recorded using a continuous-wave fNIRS system. Linear discriminant analysis (LDA) is employed for training and testing, using the data from the prefrontal, left- and right-dorsolateral prefrontal regions. For classification, eight features are tested: mean oxyhemoglobin, mean deoxyhemoglobin, skewness, kurtosis, signal slope, number of peaks, sum of peaks, and signal peak, in 0~5, 0~10, and 0~15 second time windows, respectively. The results show that the best performance for classification is achieved using mean oxyhemoglobin, the signal peak, and the sum of peaks as features. The average accuracies in the right dorsolateral prefrontal cortex (83.1, 83.4 and 84.9% in the 0~5, 0~10 and 0~15 second time windows, respectively) show that the proposed method has an effective utility for detection of drowsiness for a passive BCI. The figure below depicts 28 2-D feature spaces using a combination of two features (Subject 2). It is clear that the mean of △HbO and signal peaks provide the best data separation.

[Twenty eight combinations of 2-class feature spaces]

In this study, fNIRS is utilized to investigate the hemodynamic responses evoked in hearing the audio-stimuli of four different sound-categories, namely English-speech, non-English-speech, annoying sounds, and nature sounds, as presented in a pseudo-randomized order.

[The averaged HbOs over 18 subjects]

[Comparison of classification accuracies]

For more details, see “Decoding four different sound-categories in the auditory cortex using functional near-infrared spectroscopy,” Hearing Research, Vol., No., pp. , February 2016. (DOI: 10.1016/j.heares.2016.01.009)