Functional near-infrared spectroscopy can detect low-frequency hemodynamic oscillations in the prefrontal cortex during steady-state visual evoked potential-inducing periodic facial expression stimuli presentation
Visual Computing for Industry, Biomedicine, and Art volume 3, Article number: 28 (2020)
Brain oscillations are vital to cognitive functions, while disrupted oscillatory activity is linked to various brain disorders. Although high-frequency neural oscillations (> 1 Hz) have been extensively studied in cognition, the neural mechanisms underlying low-frequency hemodynamic oscillations (LFHO) < 1 Hz have not yet been fully explored. One way to examine oscillatory neural dynamics is to use a facial expression (FE) paradigm to induce steady-state visual evoked potentials (SSVEPs), which has been used in electroencephalography studies of high-frequency brain oscillation activity. In this study, LFHO during SSVEP-inducing periodic flickering stimuli presentation were inspected using functional near-infrared spectroscopy (fNIRS), in which hemodynamic responses in the prefrontal cortex were recorded while participants were passively viewing dynamic FEs flickering at 0.2 Hz. The fast Fourier analysis results demonstrated that the power exhibited monochronic peaks at 0.2 Hz across all channels, indicating that the periodic events successfully elicited LFHO in the prefrontal cortex. More importantly, measurement of LFHO can effectively distinguish the brain activation difference between different cognitive conditions, with happy FE presentation showing greater LFHO power than neutral FE presentation. These results demonstrate that stimuli flashing at a given frequency can induce LFHO in the prefrontal cortex, which provides new insights into the cognitive mechanisms involved in slow oscillation.
Brain oscillations are the instantiation of neuronal rhythms resulting from the dynamic reciprocity between intrinsic cellular and circuit properties , which can be ubiquitously detected across various spatial and temporal scales. Brain oscillations serve as a bridge linking single neuron activity to cognition and behaviors , in which the general functions of brain oscillators include input selection and plasticity, binding cell assemblies, consolidation, and a combination of learned information . Disrupted brain oscillations, however, are considered a hallmark of numerous brain disorders, such as schizophrenia and autism . Therefore, a thorough understanding of brain oscillations is essential to better decoding brain functions. The frequency band of brain oscillations is between 0.05 and 500 Hz . In particular, high-frequency oscillations with frequencies over 1 Hz can be detected using electroencephalography (EEG) or magnetoencephalography, indicating that different canonical frequency bands such as delta (1-4 Hz), theta (4-8 Hz), alpha (8-12 Hz), beta (12-30 Hz), and gamma (> 30 Hz) correspond to distinct cognitive functions . However, few studies have been carried out to inspect the cognitive functions associated with low-frequency oscillations, which generally fluctuate below 1 Hz. Meanwhile, low-frequency hemodynamic oscillations (LFHO) have been exploited using functional magnetic resonance imaging (fMRI) and functional near-infrared spectroscopy (fNIRS) [3,4,5], in which the low-frequency oscillation signals can be directly examined using the fast Fourier transform (FFT) of blood oxygenated (HbO) or deoxygenated (HbR) hemoglobin concentration change measurements. To date, it is still unclear whether neural activity is the major contributor to LFHO, since some studies also claim that LFHO in some frequency bands could be physiological noise . One way to capture LFHO is to explore the role of brain oscillations in cognition . For example, visual stimulus tasks can elicit steady-state visual evoked potentials (SSVEPs), which make them excellent candidates for inspecting high frequency oscillations  and LFHO .
More importantly, EEG and fMRI neuroimaging studies have been performed to inspect SSVEP in response to high-level cognitive processes, such as facial expression (FE) recognition [9,10,11,12,13,14,15,16,17,18,19]. Interestingly, previous fNIRS-based SSVEP studies mainly focused on the investigation of brain computer interfaces or brain activation in the occipital cortex [20, 21]. In contrast, the application of SSVEP in the fNIRS field [22,23,24,25] has never been used to explore how FEs relate to brain functions in the prefrontal cortex.
In this study, LFHO in the prefrontal cortex during SSVEP-inducing stimulation were inspected using fNIRS, in which the periodic flickering stimuli were FEs. In particular, FFT analysis and canonical hemodynamic response function (HRF) deconvolution were carried out to extract highly detailed spatial information on the power and phase of the HbO signals. It is hypothesized that LFHO can be induced by SSVEP-inducing flickering stimuli. It is also assumed that LFHO measures can be used to differentiate different FEs. It is expected that the present findings may contribute to a better characterization and understanding of the underlying neural mechanisms associated with LFHO.
Twenty-one college students (4 males and 17 females, mean age ± standard deviation: 22 ± 2.5 years) were recruited from the University of Macau campus. All participants were right-handed with normal vision. In addition, no history of neurologic or psychiatric disorders was reported for any participant. The present experimental tests were approved by the Biomedical Ethics Board of the University of Macau.
Stimuli and procedures
Two categories of facial stimuli (neutral and happy faces) were selected from the NimStim Face Stimulus Set . Based on the original faces, morphed faces were produced by overlaying neutral faces with happy faces, leading to 21 FEs created with a 5% gradient change (Morpheus software). For example, the 30% happy to neutral (H2N) FE indicated that the stimulation material was generated with a 70% happy face and 30% neutral one (Fig. 1).
fNIRS data were recorded while the participants were performing the FE task. The fNIRS task in Fig. 1a consisted of four blocks (conditions): neutral (condition 1), happy (condition 2), neutral to happy (N2H; condition 3) in which a neutral face changed little by little to become a happy face, and H2N (condition 4), in which a happy face gradually became a neutral face. Each block had 30 trials and each trial lasted 5.1 s. The order of block presentations was completely random, and the rest period between each of the two blocks was 30 s. Therefore, the stimulus frequency of each block was set at 0.2 Hz. Each trial started with a red cross being displayed at the center of the monitor for 1000 ms, followed by 21 FE presentations for 2100 ms and a black screen for 2000 ms (Fig. 1b). In particular, the duration for each of the 21 FE presentation was 100 ms, which included a 30 ms presentation of an FE and 70 ms presentation of a black screen to generate the flashing effect. For each block, the FE presentations for each of the 30 trials were identical.
fNIRS data acquisition and preprocessing
The fNIRS data were acquired at a sampling rate of 50 Hz using our CW6 system (Techen Inc., Milford, MA) [27,28,29,30]. Four laser sources at wavelengths of 690 nm and 830 nm and four light detectors were placed on a standard EEG cap to generate seven channels (Fig. 2), in which the spatial distance between each laser source and each detector was 3 cm. It should be noted that the fNIRS channels were located in the exact positions of the EEG electrodes: C-5 (Ch 01), E-3 (Ch 02), D-6 (Ch 03), F-4 (Ch 04), E-5 (Ch 05), E-6 (Ch 06), and F-6 (Ch 07).
The raw fNIRS data were first transformed to light density changes measured as optical density (OD) according to the modified Beer-Lambert law and then motion-corrected . Further, a low bandpass filter at 0.25 Hz and then a high bandpass filter at 0.02 Hz were used to eliminate the effect of physiological noise. The high-pass filter removes low-frequency measurement noise while the low-pass filters remove physiological noise, such as heartbeat (1 Hz) and respiratory signals (0.25-0.3 Hz). Subsequently, the data were segmented into 5.1 s epochs for each trial consisting of 1 s before and 4.1 s after trigger onset. Finally, HbO and HbR hemoglobin changes were generated using OD values, and only HbO signals were analyzed in this study because they can serve as more sensitive indicators of changes associated with regional cerebral blood flow. The grand-averaged HbO signals across all conditions and participants are provided in Fig. 3 for the seven channels.
In this study, the group-averaged HbO time series from seven channels were converted to frequency-domain signals via FFT. The FFT is usually applied to generate SSVEPs in EEG analysis, and was also used here to calculate the distributions of LFHO in the frequency domain. In addition, the signal noise ratio (SNR) of LFHO was computed here, which was equal to the power of a specific frequency divided by the averaged power of the 20 surrounding frequency bins taking the 10 nearest ones on each side [16, 32]. Further, the grand-averaged SNR time series of each channel was calculated for each condition across all participants. Importantly, the z-score of the grand-averaged SNR was generated for each condition, which was denoted as the power of a frequency that was first subtracted from the mean of 20 surrounding frequency bins and then divided by its standard deviation [16, 32].
LFHO across the prefrontal cortex
As depicted in Fig. 4, LFHO were successfully evoked with periodic stimuli presentation at a frequency of 0.2 Hz for all channels. The LFHO exhibited a very similar distribution to the SSVEP, in which grand-averaged SNR peaks were detected at the fundamental frequency of the stimuli (0.2 Hz) and its harmonics (0.4 Hz).
In addition, resting-state and task-elicited fNIRS recordings without flickering stimuli with SSVEP were also acquired in the frontal area using our CW6 system [24, 25]. The two sets of fNIRS data were also preprocessed and analyzed using the same procedure as LFHO during SSVEP-inducing periodic FE stimuli presentation. As displayed in Fig. 5, no monochronic peaks in SNR at a frequency of 0.2 Hz were identified for either fNIRS dataset. It was discovered that without external periodic stimuli, no LFHO at the dominant frequency can be elicited in the prefrontal cortex.
LFHO after deconvolution
In addition, to inspect whether LFHO were independent of neurovascular coupling, HRF deconvolution was carried out to generate the “neural level signals” (LFHO after HRF deconvolution). The HRF was produced for each channel by matching the HbO with the canonical HRF and its time derivative. The blind HRF deconvolution is capable of eliminating the effect of hemodynamic responses to a maximal extent, in which the HRF is based on the convolution of the boxcar function and the sum of two gamma functions as the canonical HRF [4, 33]. The deconvoluted HbO data before and after FFT are plotted in Figs. 6 and 7, respectively. It was discovered that LFHO were detected at 0.2 Hz. The HbO/SNR signals from all channels before and after HRF deconvolution were compared under various conditions. It was discovered that the distributions of the SNR did not exhibit significant differences after HRF deconvolution, which demonstrates that LFHO are independent of neurovascular coupling.
Meanwhile, paired t-tests were performed to examine the difference in SNR amplitude of the dominant frequency (0.2 Hz for the present study) between the neutral and happy conditions, as well as the difference between the N2H and H2N conditions. All statistical analyses were conducted using SPSS 18.0 software. The SNR maps across all channels under different conditions are depicted in Fig. 8. Paired t-tests (t15 = 2.20, p = 0.044, CI = 0.09-5.54) demonstrated that the happy condition (13.73 ± 5.38) manifested greater brain activation than the neural condition (10.87 ± 5.21). By contrast, no significant difference between the N2H and H2N conditions was found (t15 = 0.36, p = 0.73, CI = − 4.15-5.83).
The main purpose of this study was to explore whether fNIRS can detect low-frequency oscillations elicited by periodic events in the prefrontal cortex. Accordingly, exogenous flashing FEs at 0.2 Hz were utilized as periodic stimuli to induce SSVEP. The FFT results of the HbO data show that there were predominant peaks of power at 0.2 Hz across all channels in the prefrontal cortex, indicating that LFHO were successfully evoked as a new indicator of cognitive function or brain disorders. More importantly, the power amplitudes at 0.2 Hz can exhibit differences in brain activation between different cognitive states; the happy condition elicited larger power amplitudes than the neutral condition. Interestingly, previous EEG studies also demonstrated that emotional stimuli elicited larger SSVEPs than neutral stimuli [11,12,13,14,15, 19].
It is widely recognized that SSVEPs are elicited mainly in electrodes over the occipital and parietal cortex, which exhibit oscillatory components when the participant is watching flickering at a constant frequency. However, previous studies mainly focused on the investigation of high-frequency neural oscillation activity with a frequency above 1 Hz. In contrast, low-frequency oscillations (lower than 1 Hz) are seldom examined given that EEG is not an ideal tool to inspect low-frequency oscillations. This study demonstrated that fNIRS is an ideal tool to evaluate LFHO in the prefrontal cortex that are related to high-level cognition, such as processing emotional FEs. Of note, due to the relatively low spatial resolution of fNIRS, the accuracy could be improved by the simultaneous utilization of Fourier and image space analysis [34, 35].
Although fNIRS can only indirectly record neural activity, it was discovered that LFHO detected by fNIRS somehow reflect the underlying neural activity and are independent of neurovascular coupling. Given that the peak of LFHO at 0.2 Hz was induced by periodic FEs flickering at the same frequency, we infer that the high power amplitude of the happy condition might result from the underlying neutral activity rather than other physiological noise.
Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Fast Fourier transform
Functional magnetic resonance imaging
Functional near-infrared spectroscopy
Happy to neutral
Hemodynamic response function
Low-frequency hemodynamic oscillations
Neutral to happy
Signal noise ratio
Steady state visual evoked potentials
Buzsáki G, Draguhn A (2004) Neuronal oscillations in cortical networks. Science 304(5679):1926–1929. https://doi.org/10.1126/science.1099745
Wang XJ (2010) Neurophysiological and computational principles of cortical rhythms in cognition. Physiol Rev 90(3):1195–1268. https://doi.org/10.1152/physrev.00035.2008
Zuo XN, Di Martino A, Kelly C, Shehzad ZE, Gee DG, Klein DF et al (2010) The oscillating brain: complex and reliable. Neuroimage 49(2):1432–1445. https://doi.org/10.1016/j.neuroimage.2009.09.037
Lu FM, Wang YF, Zhang J, Chen HF, Yuan Z (2017) Optical mapping of the dominant frequency of brain signal oscillations in motor systems. Sci Rep 7(1):14703. https://doi.org/10.1038/s41598-017-15046-9
Wang YF, Chen W, Ye LK, Biswal BB, Yang XZ, Zou QJ et al (2018) Multiscale energy reallocation during low-frequency steady-state brain response. Hum Brain Mapp 39(5):2121–2132. https://doi.org/10.1002/hbm.23992
Klimesch W (1999) EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis. Brain Res Rev 29(2–3):169–195. https://doi.org/10.1016/S0165-0173(98)00056-3
Regan D (1989) Human brain electrophysiology: evoked potentials and evoked magnetic fields in science and medicine. Elsevier, New York
Lewis LD, Setsompop K, Rosen BR, Polimeni JR (2016) Fast fMRI can detect oscillatory neural activity in humans. Proc Natl Acad Sci U S A 113(43):E6679–E6685. https://doi.org/10.1073/pnas.1608117113
Vialatte FB, Maurice M, Dauwels J, Cichocki A (2010) Steady-state visually evoked potentials: focus on essential paradigms and future perspectives. Prog Neurobiol 90(4):418–438. https://doi.org/10.1016/j.pneurobio.2009.11.005
Norcia AM, Appelbaum LG, Ales JM, Cottereau BR, Rossion B (2015) The steady-state visual evoked potential in vision research: a review. J Vis 15(6):4. https://doi.org/10.1167/15.6.4
Wieser MJ, Miskovic V, Keil A (2016) Steady-state visual evoked potentials as a research tool in social affective neuroscience. Psychophysiology 53(12):1763–1775. https://doi.org/10.1111/psyp.12768
Wieser MJ, Keil A (2014) Fearful faces heighten the cortical representation of contextual threat. Neuroimage 86:317–325. https://doi.org/10.1016/j.neuroimage.2013.10.008
Wieser MJ, McTeague LM, Keil A (2011) Sustained preferential processing of social threat cues: bias without competition? J Cogn Neurosci 23(8):1973–1986. https://doi.org/10.1162/jocn.2010.21566
Wieser MJ, McTeague LM, Keil A (2012) Competition effects of threatening faces in social anxiety. Emotion 12(5):1050–1060. https://doi.org/10.1037/a0027069
Bekhtereva V, Craddock M, Müller MM (2015) Attentional bias to affective faces and complex IAPS images in early visual cortex follows emotional cue extraction. Neuroimage 112:254–266. https://doi.org/10.1016/j.neuroimage.2015.03.052
Dzhelyova M, Jacques C, Rossion B (2017) At a single glance: fast periodic visual stimulation uncovers the spatio-temporal dynamics of brief facial expression changes in the human brain. Cereb Cortex 27(8):4106–4123. https://doi.org/10.1093/cercor/bhw223
Foley E, Rippon G, Senior C (2018) Modulation of neural oscillatory activity during dynamic face processing. J Cogn Neurosci 30(3):338–352. https://doi.org/10.1162/jocn_a_01209
Mayes AK, Pipingas A, Silberstein RB, Johnston P (2009) Steady state visually evoked potential correlates of static and dynamic emotional face processing. Brain Topogr 22(3):145–157. https://doi.org/10.1007/s10548-009-0106-5
McTeague LM, Shumen JR, Wieser MJ, Lang PJ, Keil A (2011) Social vision: sustained perceptual enhancement of affective facial cues in social anxiety. Neuroimage 54(2):1615–1624. https://doi.org/10.1016/j.neuroimage.2010.08.080
Di Lorenzo R, Blasi A, Junge C, van den Boomen C, van Rooijen R, Kemner C (2019) Brain responses to faces and facial expressions in 5-month-olds: an fNIRS study. Front Psychol 10:1240. https://doi.org/10.3389/fpsyg.2019.01240
Tomita Y, Vialatte FB, Dreyfus G, Mitsukura Y, Bakardjian H, Cichocki A (2014) Bimodal BCI using simultaneously NIRS and EEG. IEEE Trans Biomed Eng 61(4):1274–1284. https://doi.org/10.1109/TBME.2014.2300492
He Y, Wang MY, Li DF, Yuan Z (2017) Optical mapping of brain activation during the English to Chinese and Chinese to English sight translation. Biomed Opt Express 8(12):5399–5411. https://doi.org/10.1364/BOE.8.005399
Ren HH, Wang MY, He Y, Du ZC, Zhang J, Zhang J et al (2019) A novel phase analysis method for examining fNIRS neuroimaging data associated with Chinese/English sight translation. Behav Brain Res 361:151–158. https://doi.org/10.1016/j.bbr.2018.12.032
Wang MY, Lu FM, Hu ZS, Zhang J, Yuan Z (2018) Optical mapping of prefrontal brain connectivity and activation during emotion anticipation. Behav Brain Res 350:122–128. https://doi.org/10.1016/j.bbr.2018.04.051
Wang MY, Zhang J, Lu FM, Xiang YT, Yuan Z (2018) Neuroticism and conscientiousness respectively positively and negatively correlated with the network characteristic path length in dorsal lateral prefrontal cortex: a resting-state fNIRS study. Brain Behav 8(9):e01074. https://doi.org/10.1002/brb3.1074
Tottenham N, Tanaka JW, Leon AC, McCarry T, Nurse M, Hare TA et al (2009) The NimStim set of facial expressions: judgments from untrained research participants. Psychiatry Res 168(3):242–249. https://doi.org/10.1016/j.psychres.2008.05.006
Hu ZS, Zhang J, Couto TA, Xu SY, Luan P, Yuan Z (2018) Optical mapping of brain activation and connectivity in occipitotemporal cortex during Chinese character recognition. Brain Topogr 31(6):1014–1028. https://doi.org/10.1007/s10548-018-0650-y
Ieong HFH, Yuan Z (2018) Emotion recognition and its relation to prefrontal function and network in heroin plus nicotine dependence: a pilot study. Neurophotonics 5(2):025011. https://doi.org/10.1117/1.NPh.5.2.025011
Lin XH, Sai LY, Yuan Z (2018) Detecting concealed information with fused electroencephalography and functional near-infrared spectroscopy. Neuroscience 386:284–294. https://doi.org/10.1016/j.neuroscience.2018.06.049
Yuan Z, Zhang J, Wang XD, Li CQ (2014) A systematic investigation of reflectance diffuse optical tomography using nonlinear reconstruction methods and continuous wave measurements. Biomed Opt Express 5(9):3011–3022. https://doi.org/10.1364/BOE.5.003011
Huppert TJ, Diamond SG, Franceschini MA, Boas DA (2009) HomER: a review of time-series analysis methods for near-infrared spectroscopy of the brain. Appl Opt 48(10):D280–D298. https://doi.org/10.1364/AO.48.00D280
Liu-Shuang J, Norcia AM, Rossion B (2014) An objective index of individual face discrimination in the right occipito-temporal cortex by means of fast periodic oddball stimulation. Neuropsychologia 52:57–72. https://doi.org/10.1016/j.neuropsychologia.2013.10.022
Wu GR, Liao W, Stramaglia S, Ding JR, Chen HF, Marinazzo D (2013) A blind deconvolution approach to recover effective connectivity brain networks from resting state fMRI data. Med Image Anal 17(3):365–374. https://doi.org/10.1016/j.media.2013.01.003
Wang SS, Cheng HT, Ying L, Xiao TH, Ke ZW, Zheng HR et al (2020) DeepcomplexMRI: exploiting deep residual network for fast parallel MR imaging with complex convolution. Magn Reson Imaging 68:136–147. https://doi.org/10.1016/j.mri.2020.02.002
Wang SS, Ke ZW, Cheng HT, Jia S, Ying L, Zheng HR et al (2019) DIMENSION: dynamic MR imaging with both k-space and spatial prior knowledge obtained via multi-supervised network training. NMR Biomed:e4131. https://doi.org/10.1002/nbm.4131
University of Macau, Nos. MYRG2019-00082-FHS and MYRG2018-00081-FHS; Macao Science and Technology Development Fund, No. FDCT 025/2015/A1 and FDCT 0011/2018/A1.
Consent for publication
The facial expressions were chosen from a public dataset, and the facial pictures presented in Fig. 1 were consented to use in this publication.
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Wang, MY., Yuan, A., Zhang, J. et al. Functional near-infrared spectroscopy can detect low-frequency hemodynamic oscillations in the prefrontal cortex during steady-state visual evoked potential-inducing periodic facial expression stimuli presentation. Vis. Comput. Ind. Biomed. Art 3, 28 (2020). https://doi.org/10.1186/s42492-020-00065-7
- Steady state visual evoked potentials
- Dynamic facial expressions
- Functional near-infrared spectroscopy
- Brain oscillation