参见作者原研究论文

本实验方案简略版
Jun 2021

本文章节


 

Stimulus-induced Robust Narrow-band Gamma Oscillations in Human EEG Using Cartesian Gratings
使用笛卡尔光栅在人类脑电图中刺激诱导的稳健窄带伽马振荡   

引用 收藏 提问与回复 分享您的反馈 Cited by

Abstract

Stimulus-induced narrow-band gamma oscillations (20–70 Hz) are induced in the visual areas of the brain when particular visual stimuli, such as bars, gratings, or full-screen hue, are shown to the subject. Such oscillations are modulated by higher cognitive functions, like attention, and working memory, and have been shown to be abnormal in certain neuropsychiatric disorders, such as schizophrenia, autism, and Alzheimer’s disease. However, although electroencephalogram (EEG) remains one of the most non-invasive, inexpensive, and accessible methods to record brain signals, some studies have failed to observe discernable gamma oscillations in human EEG. In this manuscript, we have described in detail a protocol to elicit robust gamma oscillations in human EEG. We believe that our protocol could help in developing non-invasive gamma-based biomarkers in human EEG, for the early detection of neuropsychiatric disorders.

Keywords: EEG (EEG), Gamma (Gamma), Stimulus-induced gamma (刺激诱导的Gamma), Artifact rejection (排除伪迹)

Background

Gamma oscillations are narrow-band oscillations (~20–70 Hz) observed in the electrical activity of the brain. These oscillations are thought to originate from excitatory-inhibitory networks in the brain (Buzsáki and Wang, 2012), and are proposed to be involved in certain higher cognitive functions, like feature binding (Gray et al., 1989), attention (Gregoriou et al., 2009; Chalk et al., 2010), and working memory (Pesaran et al., 2002). Previous studies have described these oscillations in various species, such as cat (Gray et al., 1989; Siegel and König, 2003), mouse (Colgin et al., 2009; Veit et al., 2017), macaques (Gieselmann and Thiele, 2008; Ray and Maunsell, 2010), and humans (Muthukumaraswamy and Singh, 2013; van Pelt and Fries, 2013); as well as in various brain regions, such as the hippocampus (Colgin et al., 2009), amygdala (Amir et al., 2018), the olfactory cortex (Kay, 2003), and the visual cortex (Gieselmann and Thiele, 2008; Ray and Maunsell, 2010; Shirhatti and Ray, 2018).


Specifically, in the visual cortex, these oscillations are robustly elicited by certain stimuli called gratings. Such stimulus-induced narrow-band gamma oscillations have been reported and studied in intracranial recordings (Gieselmann and Thiele, 2008; Murty et al., 2018), and human magneto-encephalography (Muthukumaraswamy and Singh, 2013; van Pelt and Fries, 2013). However, some studies (for example, Juergens et al., 1999; Orekhova et al., 2015) have failed to elicit discernable gamma oscillations in human electro-encephalography (EEG), a neurophysiological tool of great clinical relevance for its non-invasive and inexpensive nature.


Failure to observe gamma oscillations in human EEG could be due to stimulus properties. Gamma oscillations are highly dependent on the properties of gratings, such as their spatial frequency, size, and contrast. For example, the power of these oscillations increases with the size of gratings (Gieselmann and Thiele, 2008; Ray and Maunsell, 2011; Muthukumaraswamy and Singh, 2013), and decreases with the drifting speed of gratings (Orekhova et al., 2015; Murty et al., 2018). However, many studies that examined gamma oscillations in human EEG have used smaller stimuli, stimuli that were not always presented in all quadrants of the visual space (for example, Muthukumaraswamy and Singh, 2013), or stimuli that were not stationary (for example, Orekhova et al., 2015). Koch et al. (2009) reported stimulus-induced gamma oscillations in human EEG for larger stimuli, but failed to observe two distinct gamma bands frequently reported in animal studies (Kay, 2003; Colgin et al., 2009; Veit et al., 2017; Murty et al., 2018).


We have consistently reported two distinct stimulus-induced narrow-band visual gamma oscillations (slow gamma [20–34 Hz] and fast gamma [36–66 Hz]) in human EEG from subjects aged 20–85 years-old (Murty et al., 2018, 2020, 2021) that were test-retest reliable (Kumar et al., 2022). Further, we found that these oscillations were variable across subjects, and were weaker in elderly subjects (Murty et al., 2020), and subjects with prodromal/early Alzheimer’s disease (Murty et al., 2021). In these studies, we used very large (full screen) stationary Cartesian gratings that elicited robust gamma oscillations, as opposed to drifting annular gratings used by Koch et al. (2009) and Orekhova et al. (2015). Further, we used a bipolar referencing scheme for analysis, instead of the unipolar (Koch et al., 2009), common-average (Muthukumaraswamy and Singh, 2013), or left earlobe (Orekhova et al., 2015) references. We found that gamma oscillations were more saliently observed with bipolar compared to unipolar reference schemes (for example, see Figure 1 of Murty et al., 2020).


In the present article, we have described our protocol to robustly elicit gamma oscillations in human EEG in considerable detail. Further, we have recently shown that gamma oscillations recorded using the high-end research grade amplifiers (described below) can also be reproduced using low-cost and easily available amplifiers (OpenBCI, Inc.; Pattisapu and Ray, 2021). As gamma oscillations are suggested to be abnormal in certain neuropsychiatric disorders, such as schizophrenia (Tada et al., 2014; Hirano et al., 2015), autism (Uhlhaas and Singer, 2007; Wilson et al., 2007; An et al., 2018), and Alzheimer’s disease (Iaccarino et al., 2016; Palop and Mucke, 2016; Mably and Colgin, 2018; Murty et al., 2021), we believe that our protocol could help in developing low-cost non-invasive gamma-based biomarkers in human EEG, for early detection of such diseases.

Materials and Reagents

  1. Setup for subjects’ seating area (Figure 1):

    1. Custom-designed chin rest (with mounted head rest).

    2. A table for stimulus presentation monitor (mentioned in the Equipment subsection) and chin rest.

    3. A stable chair for the subject to sit.

  2. Inch tape (for measuring scalp diameter, as well as the distance between the monitor and the subject).

  3. Personal items for cleaning the gel off scalp after the experiment, like tissue paper, towel, shampoo, etc.



    Figure 1. Figure showing EEG setup in the subject area with and without the subject.

Equipment

  1. EEG acquisition system

    1. Amplifier: BrainAmp DC (Brain Products GmbH, Germany)

    2. Electrodes: 64 active electrodes (Brain Products GmbH, Germany)

    3. Electrode cap: actiCAP (Easycap GmbH, Germany)

    4. Electrode gel: SuperVisc High Viscosity Electrolyte-Gel (Easycap GmbH, Germany)

    5. Recording computer: Windows PC with EEG recording software

  2. Stimulus presentation monitor (in subject area): BenQ XL2411 (LCD, 1280×720 pixels, refresh rate: 100 Hz)

  3. Monitor calibrating device (for gamma correction): i1Display Pro (X-Rite)

  4. Eye-tracker: EyeLink 1000 (SR Research Ltd, Canada). This includes the infrared eye-tracking camera and the eye-data acquisition computer

  5. Stimulus presentation computer: iMac (Mac OS X El Capitan)

Software

  1. Stimulus presentation: custom library and app built in objective-c for Mac OS

  2. EEG recording: Brain Vision Recorder (Version 1.20.0701)

  3. Data analysis: MATLAB, with the following toolboxes/code:

    1. Chronux toolbox for signal processing

    2. Custom code (https://github.com/supratimray/TLSAEEGProjectPrograms) and EEGLAB for generating figures

Procedure

In the following sections, we describe in detail the procedure for EEG recording, artefact rejection, and data analysis used in Murty et al. (2020, 2021). These methods are built upon and have evolved from the methods that we used in Murty et al. (2018) . For minor differences in these methods, we direct the readers to the original articles.

  1. EEG room setup

    The experimenter’s rig comprised of a stimulus presentation computer, a recording computer, an analysis computer (optional), and an eye-data acquisition computer. The subject area was separated from the rig using thick curtains that also made sure that it was dark. Subjects sat on a stable chair and placed their chin on a chin rest. Their head was supported by cheek and head-rests mounted on the chin rest. We placed the gamma corrected monitor in front of and at the level of the chin rest, at a mean ± standard deviation (SD) distance of 58.1 ± 0.8 cm (range: 53.8–61.0 cm) from the subjects, according to their convenience. The monitor subtended a width of at least 52° and height of at least 30° of the visual field. The chin rest or its components never obstructed the monitor, and the entire monitor was always visible to the subjects. We calibrated the stimuli to the viewing distance for all subjects. We placed the eye-tracking camera below and at the center of the monitor. We instructed the subjects to wear spectacles, if prescribed.

  2. EEG recording

    On the day of the experiment, we requested subjects to wash their hair with shampoo prior to coming for the recording session. We also briefed the subjects about the procedure and took their written consent before starting. Electrodes were placed on the scalp according to the international 10–10 system, using the electrode cap and gel as suggested by the manufacturer. We ensured that the gel under one electrode did not come in contact with the gel under another electrode. We also ensured that electrode impedances were less than 25 KΩ while applying gel, and rejected those electrodes that could not satisfy this criterion from analysis. After this process, we seated the subjects on the subject’s chair and calibrated the eye-tracker. We then presented a set of stimuli on the stimulus presentation monitor, while the subjects fixated on a small (0.1°) fixation point on the monitor. We acquired raw EEG signals by referencing all 64 electrodes to location ‘FCz’ on the scalp (see Figure 5). We filtered these signals online between 0.016 Hz (first-order filter) and 1000 Hz (fifth-order Butterworth filter), sampled at 2,500 Hz, and digitized at 16-bit resolution (0.1 μV/bit).

  3. Stimuli

    Stimuli were static achromatic images whose luminosity varied sinusoidally across space, giving a sense of continuous, alternating black-and-white bars on the monitor. Such images are called gratings. These gratings are characterized by spatial frequency and orientation: spatial frequency is the frequency of the sinusoid (expressed as cycles per degree [cpd] of the visual field), and orientation is the direction perpendicular to the direction of the sinusoid (0° being vertical gratings). Gratings could further be varied in terms of their size (diameter) and contrast. We had observed that gratings with spatial frequency 1–4 cpd, having a diameter larger than 16° of visual field, and at 100% of contrast elicited robust slow and fast gamma oscillations in humans (Murty et al., 2018). Video 1 shows example full-screen stimuli (varied in spatial frequency and orientation) at 100% contrast.


    Video 1. Video showing example gratings and example trials.

    Each trial began with the appearance of a fixation point at the centre of the stimulus presentation screen. Within a trial, a series of gratings were presented (2 to 3 per trial) for 800 ms, with an interstimulus interval of 700 ms. Gratings were achromatic sinusoidal Cartesian gratings, whose spatial frequency was randomly chosen from 1, 2, and 4 cpd. The orientation of the gratings was randomly chosen from 0°, 45°, 90°, and 135°. The fixation point was presented on the screen throughout the trial, and disappeared at the end of the trial.


  4. Stimulus presentation

    Details of stimulus presentation could be found in Murty et al. (2018, 2020, 2021). Example trials are shown in Video 1, and a video of a subject performing the task is shown in Video 2. Typically, the subjects fixated on the fixation point for 1,000 ms at the start of each trial. After this initial fixation period, we presented a series of gratings (2 to 3 per trial) for 800 ms, with an interstimulus interval of 700 ms. After the last stimulus of the trial was presented, the fixation point disappeared, and the subjects were allowed to rest their eyes by breaking fixation, while keeping their head stable. We performed each experiment in 1–2 blocks in a single session, according to the comfort of the subject. We gave breaks between blocks, during which the subjects could relax by removing their head off the chin rest. However, the distance of the chin rest from the monitor was kept constant between blocks. For a few subjects reported in Murty et al. (2018, 2020), we performed the experiments in two sessions according to the comfort of the subjects. For each session, we repeated the entire procedure of eye-tracker and stimulus calibration for monitor-to-chin rest distance.


    Video 2. Video of a subject performing the task.

    Subjects had to fixate on the fixation point throughout the trial. The trial was aborted if they broke fixation during the trial. After the trial ended or aborted, they were allowed to rest their eyes by breaking fixation, but keeping their head stable. Sounds were played through in-built speakers on iMac (the stimulus presentation computer) that alerted the subjects at the start of the trial, as well as when the trial ended, or when they broke fixation during the trial.


  5. Data segmentation

    We first segmented EEG data between -1 to 2.2768 seconds around each stimulus onset time (total: 8192 points), and generated a N x 8192 data matrix for each electrode, where N is the number of stimulus repeats. Each segment is abbreviated as a “repeat”.

  6. Artifact rejection

    We wanted to find a common set of bad repeats for all the electrodes, which would allow us to also do connectivity/causality analysis. In many cases, a small set of electrodes had a large number of bad repeats, so including those electrodes drastically reduced the total number of repeats for all electrodes. Therefore, we designed a fully automated artifact rejection pipeline to (i) find a set of bad electrodes, and (ii) a set of common bad repeats in the remaining good electrodes.

    As mentioned above, we rejected data offline from those electrodes whose impedance was more than 25 KΩ. We further rejected repeats with fixation breaks (defined as eye-blinks or shifts in eye-position outside a square window of width 5º, centered on the fixation spot) during 0.5s–0.75s of stimulus onset. We preprocessed the recorded data as follows (also described in Murty et al., 2020):

    1. For each unipolar electrode (Figure 2): We considered data for each electrode into smaller segments from -500 ms to 750 ms of stimulus onset. We then applied trial-wise thresholds in both time and frequency domain, as follows:

      1. For raw waveforms (time-domain): We first high-pass filtered the signal for each repeat at 1.6 Hz, to eliminate slow trends if any. We filtered the signal only for the purpose of artefact rejection, but not for final data analysis. Any stimulus repeat for which the filtered waveform deviated by six times the standard deviation (SD) from the mean at any time bin was considered as a bad repeat for that electrode.

      2. For Power Spectral Density (PSD; frequency-domain): We estimated multi-tapered PSDs from 0 to 200 Hz for each trial separately, with five tapers (time-bandwidth product or TW=3), using Chronux toolbox [(Mitra and Bokil, 2008), http://chronux.org/, RRID:SCR_005547]. Any stimulus repeat, for which the PSD deviated by 6 SD from the mean at any frequency point, was considered as a bad repeat for that electrode.



      Figure 2. Flowchart for separating bad repeats for each unipolar electrode.


    2. Common set of bad repeats (Figure 3): We next created a common set of bad repeats across all 64 unipolar electrodes, as follows:

      1. We first discarded those electrodes that had more than 30% of all repeats marked as bad (from step a).

      2. We then created a common set of bad repeats, by including in it any repeat (that was marked as bad in step a) in any of the ten unipolar electrodes used for analysis (P3, P1, P2, P4, PO3, POz, PO4, O1, Oz, and O2).

      3. Next, we included in this set any repeat (that was marked as bad in step a), if it occurred in more than 10% of the total number of remaining electrodes. This was the final set of bad repeats that we discarded, so the rest of the stimulus repeats were used for analysis.



      Figure 3. Flowchart for creating a common set of bad repeats across all analyzable unipolar electrodes.


    3. Further rejection of electrodes (Figure 4): As mentioned above, we discarded those electrodes that had impedance more than 25 KΩ, or those that had more than 30% of all repeats marked as bad. Occasionally, we observed electrodes for which the PSD tended to flatten out (which could happen if it hit a noise floor, for example). We therefore added another pipeline to reject such electrodes. To this end, we calculated PSD slopes for each unipolar electrode, as follows:

      1. We estimated PSDs for each analyzable repeat using one taper (TW=1) and averaged the resulting PSDs.

      2. We fitted a power-law function, P(f)=A.f to these trial-averaged PSDs, where A (scaling factor), and β (slope) are free parameters obtained using least square minimization (using the fminsearch function in MATLAB), and P is the PSD across frequencies f∈[56 84] Hz.

      We discarded those electrodes that had slopes less than 0. We did this separately for each block.



      Figure 4. Flowchart for discarding bad unipolar electrodes.


    For any block, if there were no analyzable bipolar electrodes (see Data Analysis subsection below), we discarded that block for that subject, and pooled the data across the rest of the blocks for data analysis. For minor differences in rejection of blocks for Murty et al. (2020, 2021), kindly refer to the respective articles.

    Data analysis

    1. Selection of electrodes

      We re-referenced data at each electrode offline to its neighboring electrodes (a reference scheme called bipolar reference), as shown in Figure 5 (Murty et al., 2020). This was done because we found that gamma oscillations were much more prominent in the bipolar referencing scheme (see Figure 1 of Murty et al., 2020). Specifically, we considered the following nine bipolar electrodes for analysis: PO3-P1, PO3-P3, POz-PO3, PO4-P2, PO4-P4, POz-PO4, Oz-POz, Oz-O1, and Oz-O2. We discarded a bipolar electrode if either of its constituting unipolar electrodes was marked bad during artifact rejection.



      Figure 5. Schematic showing bipolar montage.

      The colored electrodes are physical electrodes placed on the scalp, as per the International 10-10 System (Easycap GmbH, Germany). These electrodes are referenced online to FCz (unipolar reference scheme). Ground is at AFz. Data from adjacent physical electrodes are referenced offline with respect to each other (bipolar reference scheme). Bipolar electrodes are shown in red. Each bipolar electrode is virtually placed in between the two constituent unipolar electrodes. We analyzed 112 such electrodes in our studies.


    2. Analysis of EEG data

      We analyzed all data using custom codes written in MATLAB (The MathWorks, Inc., RRID:SCR_001622), as described in Murty et al. (2020) . We computed PSD and the time-frequency power spectrograms using a multi-taper method with a single DPSS taper (TW=1), using the Chronux toolbox as described below. We averaged raw PSDs and raw spectrograms across all analyzable stimulus repeats for the final analysis. Further, we generated scalp maps using the topoplot.m function of EEGLAB toolbox (Delorme and Makeig, 2004, RRID:SCR_007292), modified to show each electrode as a colored disc.

      1. Computation of raw PSD and change in PSD: We calculated power at each frequency (at a resolution of 2 Hz) separately for the baseline period (−500 ms and 0 ms of stimulus onset), and stimulus period (250 ms and 750 ms, avoiding stimulus onset-related transients between 0 and 250 ms). For change in PSD from baseline to stimulus periods (expressed in dB), we used the following equation:



        where P(f) is the power at the frequency f. Figure 6a shows the raw PSD (black line) for stimulus (solid line), and baseline (dashed line) periods, as well as the change in PSD (blue solid line) for an example subject. Discernible gamma oscillations are observed as ‘bumps’ in 20-34 Hz (slow gamma), and 36-66 Hz (fast gamma) frequency ranges.

      2. Computation of raw and change in time-frequency power spectrograms: We estimated raw time-frequency power spectrograms during -500 ms and 1,200 ms of stimulus onset. We chose a moving window of 250 ms, and a step size of 25 ms, giving a frequency resolution of 4 Hz. For computation of change in the power spectrogram (expressed in dB), we subtracted the power averaged across the baseline period of the raw spectrogram from the entire raw spectrogram, for each frequency. In other words, we used the following formula:



        where Sbl (f) is the baseline power at frequency (f) (averaged from -0.5 to 0 s). Figure 6b shows a raw spectrogram (top row) and a change in power spectrogram (bottom row), for the same example subject as in Figure 6a. Both slow and fast gamma oscillations are robustly observed in the change in power spectrogram.

      3. Computation of raw power in different frequency bands: To estimate raw power in frequency bands of interest (i.e., alpha, slow gamma and fast gamma) during baseline and stimulus periods, we summed the power at individual frequency bins in the respective frequency bands as estimated in the PSDs. In other words, we used the following equation:



        where P(f) is the power at the frequency f, and a and b are the lower and upper limits of the frequency band of interest.

      4. Computation of stimulus-induced change in power in different frequency bands: To estimate the change in power during the stimulus period compared to the baseline period, we used the following equation for each frequency band:




      Figure 6. Spectra and spectrograms in an example subject.

      a) Plot showing raw PSD (left axis) in the stimulus (black solid line) and baseline (black dashed line) periods, averaged across nine bipolar electrodes (PO3-P1, PO3-P3, POz-PO3, PO4-P2, PO4-P4, Poz-PO4, Oz-Poz, Oz-O1, and Oz-O2), and expressed on a log10 scale. On the axis on the right side, the change in PSD (solid blue line) from stimulus to baseline periods is shown on a dB scale. Slow gamma (20–34 Hz) and fast gamma (36–66 Hz) bands are shown in vertical violet and orange lines respectively. b) A plot showing raw (top panel) and change in power time-frequency spectrograms (bottom panel), for the same data as in a. Gamma bands are shown in solid white (slow gamma), and dashed (fast gamma) lines. Vertical black dashed lines indicate the time when stimulus was presented to the subjects (0–0.8 s), whereas red vertical dashed lines indicate the stimulus period considered for spectral analysis (0.25–0.75 s).


    Codes for these analyses are available at: https://github.com/supratimray/TLSAEEGProjectPrograms.

    Codes for artifact rejection pipeline: https://github.com/supratimray/CommonPrograms under ReadData/findBadTrialsEEG.

    Acknowledgments

    This work was supported by Tata Trusts Grant, Wellcome Trust/DBT India Alliance (Intermediate fellowship 500145/Z/09/Z and Senior fellowship IA/S/18/2/504003 to SR), and DBT-IISc Partnership Programme (to SR). The original research article related to this protocol is as follows:

    Murty, D. V. P. S., Manikandan, K., Kumar, W. S., Ramesh, R. G., Purokayastha, S., Nagendra, B., Ml, A., Balakrishnan, A., Javali, M., Rao, N. P. and Ray, S. (2021). Stimulus-induced gamma rhythms are weaker in human elderly with mild cognitive impairment and Alzheimer's disease. Elife 10: e61666.

    Competing interests

    The authors declare no competing financial interests.

    Ethics

    All subjects participated in our studies voluntarily and were monetarily compensated for their time and effort. We obtained informed consent from all subjects before the experiment. The Institute Human Ethics Committees of Indian Institute of Science, NIMHANS, and M S Ramaiah Hospital, Bangalore approved all procedures.

    References

    1. Amir, A., Headley, D. B., Lee, S. C., Haufler, D. and Pare, D. (2018). Vigilance-Associated Gamma Oscillations Coordinate the Ensemble Activity of Basolateral Amygdala Neurons. Neuron 97(3): 656-669 e657.
    2. An, K. M., Ikeda, T., Yoshimura, Y., Hasegawa, C., Saito, D. N., Kumazaki, H., Hirosawa, T., Minabe, Y. and Kikuchi, M. (2018). Altered Gamma Oscillations during Motor Control in Children with Autism Spectrum Disorder. J Neurosci 38(36): 7878-7886.
    3. Buzsáki, G. and Wang, X. J. (2012). Mechanisms of gamma oscillations. Annu Rev Neurosci 35: 203-225.
    4. Chalk, M., Herrero, J. L., Gieselmann, M. A., Delicato, L. S., Gotthardt, S. and Thiele, A. (2010). Attention reduces stimulus-driven gamma frequency oscillations and spike field coherence in V1. Neuron 66(1): 114-125.
    5. Colgin, L. L., Denninger, T., Fyhn, M., Hafting, T., Bonnevie, T., Jensen, O., Moser, M. B. and Moser, E. I. (2009). Frequency of gamma oscillations routes flow of information in the hippocampus. Nature 462(7271): 353-357.
    6. Gieselmann, M. A. and Thiele, A. (2008). Comparison of spatial integration and surround suppression characteristics in spiking activity and the local field potential in macaque V1. Eur J Neurosci 28(3): 447-459.
    7. Gray, C. M., Konig, P., Engel, A. K. and Singer, W. (1989). Oscillatory responses in cat visual cortex exhibit inter-columnar synchronization which reflects global stimulus properties. Nature 338(6213): 334-337.
    8. Gregoriou, G. G., Gotts, S. J., Zhou, H. and Desimone, R. (2009). High-frequency, long-range coupling between prefrontal and visual cortex during attention. Science 324(5931): 1207-1210.
    9. Hirano, Y., Oribe, N., Kanba, S., Onitsuka, T., Nestor, P. G. and Spencer, K. M. (2015). Spontaneous Gamma Activity in Schizophrenia. JAMA Psychiatry 72(8): 813-821.
    10. Iaccarino, H. F., Singer, A. C., Martorell, A. J., Rudenko, A., Gao, F., Gillingham, T. Z., Mathys, H., Seo, J., Kritskiy, O., Abdurrob, F., et al. (2016). Gamma frequency entrainment attenuates amyloid load and modifies microglia. Nature 540(7632): 230-235.
    11. Juergens, E., Guettler, A. and Eckhorn, R. (1999). Visual stimulation elicits locked and induced gamma oscillations in monkey intracortical- and EEG-potentials, but not in human EEG. Exp Brain Res 129(2): 247-259.
    12. Kay, L. M. (2003). Two species of gamma oscillations in the olfactory bulb: dependence on behavioral state and synaptic interactions. J Integr Neurosci 2(1): 31-44.
    13. Koch, S. P., Werner, P., Steinbrink, J., Fries, P. and Obrig, H. (2009). Stimulus-induced and state-dependent sustained gamma activity is tightly coupled to the hemodynamic response in humans. J Neurosci 29(44): 13962-13970.
    14. Kumar, W. S., Manikandan, K., Murty, D. V. P. S., Ramesh, R. G., Purokayastha, S., Javali, M., Rao, N. P. and Ray, S. (2021). Stimulus-induced narrowband gamma oscillations are test-retest reliable in healthy elderly in human EEG. bioRxiv: 2021.2007.2006.451226.
    15. Mably, A. J. and Colgin, L. L. (2018). Gamma oscillations in cognitive disorders. Curr Opin Neurobiol 52: 182-187.
    16. Mitra, P. and Bokil, H. (2008). Observed brain dynamics. Oxford , New York: Oxford University Press.
    17. Murty, D. V. P. S., Manikandan, K., Kumar, W. S., Ramesh, R. G., Purokayastha, S., Nagendra, B., Ml, A., Balakrishnan, A., Javali, M., Rao, N. P. and Ray, S. (2021). Stimulus-induced gamma rhythms are weaker in human elderly with mild cognitive impairment and Alzheimer's disease. Elife 10: e61666.
    18. Murty, D. V. P. S., Manikandan, K., Kumar, W. S., Ramesh, R. G., Purokayastha, S., Javali, M., Rao, N. P. and Ray, S. (2020). Gamma oscillations weaken with age in healthy elderly in human EEG. Neuroimage 215: 116826.
    19. Murty, D. V. P. S., Shirhatti, V., Ravishankar, P. and Ray, S. (2018). Large Visual Stimuli Induce Two Distinct Gamma Oscillations in Primate Visual Cortex. J Neurosci 38(11): 2730-2744.
    20. Muthukumaraswamy, S. D. and Singh, K. D. (2013). Visual gamma oscillations: the effects of stimulus type, visual field coverage and stimulus motion on MEG and EEG recordings. Neuroimage 69: 223-230.
    21. Orekhova, E. V., Butorina, A. V., Sysoeva, O. V., Prokofyev, A. O., Nikolaeva, A. Y. and Stroganova, T. A. (2015). Frequency of gamma oscillations in humans is modulated by velocity of visual motion. Journal of Neurophysiology 114(1): 244-255.
    22. Palop, J. J. and Mucke, L. (2016). Network abnormalities and interneuron dysfunction in Alzheimer disease. Nat Rev Neurosci 17(12): 777-792.
    23. Pattisapu, S. and Ray, S. (2021). Stimulus-induced narrow-band gamma oscillations in humans can be recorded using open-hardware low-cost EEG amplifier. doi: https://doi.org/10.1101/2021.11.16.468841.
    24. Pesaran, B., Pezaris, J. S., Sahani, M., Mitra, P. P. and Andersen, R. A. (2002). Temporal structure in neuronal activity during working memory in macaque parietal cortex. Nat Neurosci 5(8): 805-811.
    25. Ray, S. and Maunsell, J. H. (2010). Differences in gamma frequencies across visual cortex restrict their possible use in computation. Neuron 67(5): 885-896.
    26. Ray, S. and Maunsell, J. H. (2011). Different origins of gamma rhythm and high-gamma activity in macaque visual cortex. PLoS Biol 9(4): e1000610.
    27. Shirhatti, V. and Ray, S. (2018). Long-wavelength (reddish) hues induce unusually large gamma oscillations in the primate primary visual cortex. Proc Natl Acad Sci U S A 115(17): 4489-4494.
    28. Siegel, M. and König, P. (2003). A functional gamma-band defined by stimulus-dependent synchronization in area 18 of awake behaving cats. J Neurosci 23(10): 4251-4260.
    29. Tada, M., Nagai, T., Kirihara, K., Koike, S., Suga, M., Araki, T., Kobayashi, T. and Kasai, K. (2016). Differential Alterations of Auditory Gamma Oscillatory Responses Between Pre-Onset High-Risk Individuals and First-Episode Schizophrenia. Cereb Cortex 26(3): 1027-1035.
    30. Uhlhaas, P. J. and Singer, W. (2007). What do disturbances in neural synchrony tell us about autism? Biol Psychiatry 62(3): 190-191.
    31. van Pelt, S. and Fries, P. (2013). Visual stimulus eccentricity affects human gamma peak frequency. Neuroimage 78: 439-447.
    32. Veit, J., Hakim, R., Jadi, M. P., Sejnowski, T. J. and Adesnik, H. (2017). Cortical gamma band synchronization through somatostatin interneurons. Nature Neuroscience 20(7): 951-959.
    33. Wilson, T. W., Rojas, D. C., Reite, M. L., Teale, P. D. and Rogers, S. J. (2007). Children and adolescents with autism exhibit reduced MEG steady-state gamma responses. Biol Psychiatry 62(3): 192-197.

简介

当向受试者显示特定的视觉刺激(例如条形、光栅或全屏色调)时,会在大脑的视觉区域中诱发刺激引起的窄带伽马振荡(20-70 Hz)。 这种振荡受高级认知功能(如注意力和工作记忆)的调节,并且已被证明在某些神经精神疾病(如精神分裂症、自闭症和阿尔茨海默病)中是异常的。 然而,尽管脑电图 (EEG) 仍然是记录脑信号的最无创、廉价和可访问的方法之一,但一些研究未能观察到人类脑电图中可辨别的伽马振荡。 在这份手稿中,我们详细描述了一种在人类脑电图中引发鲁棒伽马振荡的协议。 我们相信我们的方案可以帮助开发人类脑电图中基于伽马的非侵入性生物标志物,以早期发现神经精神疾病。


背景

在大脑电活动中观察到的窄带振荡(~20 – 70 Hz)。这些振荡被认为起源于大脑中的兴奋抑制网络(Buzsáki and Wang, 2012) ,并被认为与某些高级认知功能有关,如特征绑定(Gray et al. , 1989) 、注意力(Gregoriou et等人,2009;Chalk等人,2010)和工作记忆(Pesaran等人,2002) 。以前的研究已经描述了各种物种的这些振荡,例如猫(Gray et al. , 1989; Siegel and König, 2003) 、老鼠(Colgin et al. , 2009; Veit et al. , 2017) 、猕猴(Gieselmann and Thiele ) , 2008; Ray 和 Maunsell, 2010)和人类(Muthukumaraswamy 和 Singh, 2013; van Pelt 和 Fries, 2013) ;以及不同的大脑区域,例如海马体(Colgin et al. , 2009) 、杏仁核(Amir et al. , 2018) 、嗅觉皮层(Kay, 2003)和视觉皮层(Gieselmann and Thiele, 2008) ;Ray 和 Maunsell,2010 年;Shirhatti 和 Ray,2018 年) 。
具体来说,在视觉皮层中,这些振荡是由称为光栅的某些刺激强烈引发的。这种刺激引起的窄带伽马振荡已在颅内记录(Gieselmann 和 Thiele,2008 年;Murty等人,2018 年)和人类脑磁图(Muthukumaraswamy 和 Singh,2013 年;van Pelt 和 Fries,2013年)中得到报道和研究) .然而,一些研究(例如,Juergens等人,1999 年;Orekhova等人,2015 年)未能在人类脑电图 (EEG) 中引发可辨别的伽马振荡,这是一种对其非侵入性具有重要临床意义的神经生理学工具和便宜的性质。
未能观察到人类脑电图中的伽马振荡可能是由于刺激特性。伽马振荡高度依赖于光栅的特性,例如 它们的空间频率、大小和对比度。例如,这些振荡的功率s随着光栅尺寸的增加而增加(Gieselmann 和 Thiele,2008;Ray 和 Maunsell,2011;Muthukumaraswamy 和 Singh,2013) ,并且随着光栅的漂移速度而减小(Orekhova等人, 2015 年;穆尔蒂等人,2018 年) 。然而,许多检查人类脑电图伽马振荡的研究使用了较小的刺激,这些刺激并不总是出现在视觉空间的所有象限中(例如,Muthukumaraswamy 和 Singh,2013) ,或者不是静止的刺激(例如, Orekhova等人,2015 年) 。科赫等人。 (2009 年)报道了刺激诱导的人类脑电图中较大刺激的伽马振荡,但未能观察到动物研究中经常报告的两个不同的伽马带(Kay,2003;Colgin等人,2009;Veit等人,2017;Murty等人)等,2018) 。
我们一直在 20 – 85 岁受试者的人类脑电图中报告了两种不同的刺激引起的窄带视觉伽马振荡(慢伽马 [20 – 34 赫兹] 和快速伽马 [36 – 66 赫兹]) (Murty等人. , 2018, 2020, 2021)是重测可靠的(Kumar et al ., 2022) 。此外,我们发现这些振荡在受试者之间是可变的,并且在老年受试者(Murty等人,2020 年)和患有前驱/早期阿尔茨海默病的受试者(Murty等人,2021 年)中较弱。在这些研究中,我们使用了非常大(全屏)的固定笛卡尔光栅,它引发了强大的伽马振荡,而不是 Koch等人使用的漂移环形光栅。 (2009)和 Orekhova等人。 (2015 年) 。此外,我们使用双极参考方案进行分析,而不是单极参考方案(Koch et al ., 2009) 、共同平均值(Muthukumaraswamy 和 Singh, 2013)或左耳垂参考(Orekhova et al ., 2015) 。我们发现,与单极参考方案相比,双极的伽马振荡更明显(例如,参见Murty等人的图 ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"zy7JT7ZH","properties":{"formattedCitation":"(2020)","plainCitation":"(2020)","noteIndex":0},"citationItems":[{"id":"2CLj4oSA/Xuv1PpT8","uris":["http://zotero.org/users/6641246/items/4AKEFU3V"],"uri":["http://zotero.org/users/6641246/items/4AKEFU3V"],"itemData":{"id":"bqdsYT1j/p7R9Og4X","type":"article-journal","abstract":"Gamma rhythms (~20–70 Hz) are abnormal in mental disorders such as autism and schizophrenia in humans, and Alzheimer’s disease (AD) models in rodents. However, the effect of normal aging on these oscillations is unknown, especially for elderly subjects in whom AD is most prevalent. In a first large-scale (236 subjects; 104 females) electroencephalogram (EEG) study on gamma oscillations in elderly subjects (aged 50–88 years), we presented full-screen visual Cartesian gratings that induced two distinct gamma oscillations (slow: 20–34 Hz and fast: 36–66 Hz). Power decreased with age for gamma, but not alpha (8–12 Hz). Reduction was more salient for fast gamma than slow. Center frequency also decreased with age for both gamma rhythms. The results were independent of microsaccades, pupillary reactivity to stimulus, and variations in power spectral density with age. Steady-state visual evoked potentials (SSVEPs) at 32 Hz also reduced with age. These results are crucial for developing gamma/SSVEP-based biomarkers of cognitive decline in elderly.","container-title":"NeuroImage","DOI":"10.1016/j.neuroimage.2020.116826","ISSN":"1053-8119","journalAbbreviation":"NeuroImage","language":"en","page":"116826","source":"ScienceDirect","title":"Gamma oscillations weaken with age in healthy elderly in human EEG","volume":"215","author":[{"family":"Murty","given":"Dinavahi V. P. S."},{"family":"Manikandan","given":"Keerthana"},{"family":"Kumar","given":"Wupadrasta Santosh"},{"family":"Ramesh","given":"Ranjini Garani"},{"family":"Purokayastha","given":"Simran"},{"family":"Javali","given":"Mahendra"},{"family":"Rao","given":"Naren Prahalada"},{"family":"Ray","given":"Supratim"}],"issued":{"date-parts":[["2020",7,15]]}},"suppress-author":true}],"schema":"https://github.com/citation-style-language/schema/raw/master/csl-citation.json"} 1,2020 )。
在本文中,我们已经相当详细地描述了我们在人类脑电图中稳健地引发伽马振荡的协议。此外,我们最近表明,使用高端研究级放大器(如下所述)记录的伽马振荡也可以使用低成本且易于获得的放大器(OpenBCI,Inc.;Pattisapu 和 Ray,2021)再现。 由于伽马振荡被认为在某些神经精神疾病中是异常的,例如精神分裂症(Tada等人,2014;Hirano等人,2015) 、自闭症(Uhlhaas 和 Singer,2007;Wilson等人,2007;An等人) . , 2018)和阿尔茨海默病(Iaccarino et al. , 2016; Palop and Mucke, 2016; Mable and Colgin, 2018; Murty et al. , 2021) ,我们相信我们的协议可以帮助开发低成本的非人类脑电图中基于伽马的侵入性生物标志物,用于早期检测此类疾病。

关键字:EEG, Gamma, 刺激诱导的Gamma, 排除伪迹

材料和试剂
受试者座位区设置(图 1):
定制设计的下巴托(带安装的头托)。
用于刺激呈现监视器(在“设备”小节中提到)和下巴托的表格。
供受试者坐的稳定椅子。
英寸胶带(用于测量头皮直径,以及显示器与对象之间的距离)。
实验后用于清洁头皮凝胶的个人物品,如纸巾、毛巾、洗发水等。


 


图 1. 显示主题区域中的 EEG 设置的图,其中包含和不包含主题。




设备


脑电采集系统
放大器:BrainAmp DC(Brain Products GmbH,德国)
电极:64 个有源电极(Brain Products GmbH,德国)
电极帽:actiCAP(Easycap GmbH,德国)
电极凝胶:SuperVisc High Viscosity Electrolyte-Gel(Easycap GmbH,德国)
记录计算机:带有EEG记录软件的Windows PC
刺激呈现监视器(在主题区域内):BenQ XL2411(LCD,1280 × 720 像素,刷新率:100 Hz)
显示器校准设备(用于伽马校正):i1Display Pro (X-Rite)
眼动仪:EyeLink 1000(加拿大 SR Research Ltd)。这包括红外眼动追踪相机和眼数据采集计算机
刺激演示计算机:iMac (Mac OS X El Capitan)




软件


Stimulus 演示: c自定义库和内置于 Mac OS 的 Objective-C 应用程序
EEG 记录:Brain Vision Recorder (版本 1.20.0701)
数据分析:MATLAB,带有以下工具箱/代码:
用于信号处理的 Chronux 工具箱
用于生成图形的自定义代码 ( https://github.com/supratimray/TLSAEEGProjectPrograms ) 和 EEGLAB




程序


在以下部分中,我们详细描述了 Murty等人使用的 EEG 记录、伪影剔除和数据分析的过程。 (2020 年,2021 年) 。这些方法是基于我们在 Murty等人中使用的方法发展而来的。 (2018 年) 。对于这些方法的细微差别,我们将读者引导至原始文章。


脑电图室设置


实验者的装备由刺激呈现计算机、记录计算机、分析计算机(可选)和眼睛数据采集计算机组成。使用厚窗帘将主题区域与钻机隔开,同时确保它是黑暗的。受试者坐在稳定的椅子上,将下巴放在下巴托上。他们的头部由脸颊支撑,头枕安装在下巴托上。根据他们的方便,我们将伽马校正监视器放置在下巴托的前面和水平处,距离受试者 58.1 ± 0.8 厘米(范围:53.8-61.0 厘米)的平均 ± 标准偏差 (SD) 距离。监视器对着视野的至少 52 °的宽度和至少 30 °的高度。下巴托或其组件从未阻碍显示器,并且整个显示器始终对受试者可见。我们将刺激校准到所有受试者的观看距离。我们将眼动追踪摄像头放置在显示器下方和中央。如果有规定,我们指示受试者戴眼镜。


脑电图记录


在实验当天,我们要求受试者在来录音之前用洗发水洗头。我们还向受试者简要介绍了该程序,并在开始前获得了他们的书面同意。根据国际 10-10 系统,使用制造商建议的电极帽和凝胶将电极放置在头皮上。我们确保一个电极下的凝胶不会与另一个电极下的凝胶接触。我们还确保在应用凝胶时电极阻抗小于 25 KΩ ,并从分析中剔除那些不能满足该标准的电极。在此过程之后,我们将受试者坐在受试者的椅子上并校准眼动仪。然后,我们在刺激呈现监视器上呈现一组刺激,而受试者注视着监视器上的一个小(0.1 ° )注视点。我们通过将所有 64 个电极引用到头皮上的位置“FCz”来获取原始 EEG 信号(参见图 5)。我们在 0.016 Hz(一阶滤波器)和 1000 Hz(五阶巴特沃斯滤波器)之间在线过滤这些信号,以 2,500 Hz 采样,并以 16 位分辨率(0.1 μV/位)进行数字化。


刺激


刺激物是静态消色差图像,其亮度在空间中呈正弦变化,在监视器上给人一种连续、交替的黑白条的感觉。这样的图像称为光栅。这些光栅的特征在于空间频率和方向:空间频率是正弦波的频率(表示为视野每度的周期 [cpd]),方向是垂直于正弦波方向的方向(0 °表示垂直光栅)。光栅的尺寸(直径)和对比度可以进一步变化。我们观察到,空间频率为 1-4 cpd、直径大于 16 °视野且对比度为 100% 的光栅在人类中引发了稳健的慢速和快速伽马振荡(Murty等人,2018 年) 。视频 1 显示了 100% 对比度的示例全屏刺激(空间频率和方向不同)。


 


视频 1. 显示示例光栅和示例试验的视频。 
每次试验都以刺激呈现屏幕中心出现一个注视点开始。在一次试验中,一系列光栅(每次试验 2 到 3 个)呈现 800 毫秒,刺激间隔为 700 毫秒。光栅是消色差正弦笛卡尔光栅,其空间频率从 1、2 和 4 cpd 中随机选择。光栅的方向从0 ° 、45 ° 、90 °和135 °中随机选择。在整个试验过程中,注视点一直显示在屏幕上,并在试验结束时消失。


刺激呈现


刺激呈现的细节可以在 Murty等人中找到。 (2018 年、2020 年、2021 年) 。示例试验显示在视频 1 中,执行任务的受试者的视频显示在视频 2 中。通常,受试者在每次试验开始时会在注视点上注视 1,000 毫秒。在这个初始固定期之后,我们展示了一系列光栅(每次试验 2 到 3 个),持续 800 毫秒,刺激间隔为 700 毫秒。在试验的最后一次刺激出现后,注视点消失,让受试者通过打破注视来休息眼睛,同时保持头部稳定。根据受试者的舒适度,我们在单个会话中以 1-2 个块进行每个实验。我们在块之间进行了休息,在此期间,受试者可以通过将头部从下巴支架上移开来放松。然而,下巴托与显示器的距离在块之间保持不变。对于 Murty等人报道的一些主题。 (2018, 2020) ,我们根据受试者的舒适度分两次进行实验。对于每个会话,我们重复了眼动仪和刺激校准的整个过程,以测量显示器到下巴的静止距离。




 


视频 2. 主题执行任务的视频。
在整个试验过程中,受试者必须注视着注视点。如果他们在试验期间破坏了固定,则试验中止。在试验结束或中止后,他们可以通过打破固定来休息眼睛,但要保持头部稳定。声音通过 iMac(刺激演示计算机)上的内置扬声器播放,在试验开始时、试验结束时或试验期间他们打破固定时提醒受试者。


数据分割


)周围的 -1 到 2.2768 秒之间分割 EEG 数据,并为每个电极生成一个 N x 8192 数据矩阵,其中 N 是刺激重复的次数。每个片段都缩写为“重复”。


工件拒绝


我们想为所有电极找到一组常见的错误重复,这将使我们还可以进行连通性/因果关系分析。在许多情况下,一小组电极有大量的错误重复,因此包括这些电极会大大减少所有电极的重复总数。因此,我们设计了一个全自动的伪影剔除流程,以 (i) 找到一组坏电极,以及 (ii) 在剩余的好电极中找到一组常见的坏重复。
如上所述,我们拒绝了来自阻抗超过 25 KΩ 的电极的离线数据。我们进一步拒绝了在刺激开始的 0.5 秒到 0.75 秒内出现注视中断的重复(定义为在宽度为 5º 的方形窗口之外的眨眼或眼睛位置的变化,以注视点为中心)。我们对记录的数据进行了如下预处理(也在 Murty等人,2020 年进行了描述) :
对于每个单极电极(图 2):我们将每个电极的数据考虑到从 -500 ms 到 750 ms 的刺激开始的较小部分。然后,我们在时域和频域中应用了试验阈值,如下所示:
对于原始波形(时域):我们首先以 1.6 Hz 的频率对每次重复的信号进行高通滤波,以消除缓慢趋势(如果有)。我们过滤信号只是为了去除伪影,而不是为了最终的数据分析。任何刺激重复,其滤波波形在任何时间 bin 中偏离平均值的标准偏差 (SD) 六倍,都被认为是该电极的不良重复。
对于功率谱密度(PSD;频域):我们使用 Chronux 工具箱 [ (Mitra 和 Bokil , 2008) , http://chronux.org/, RRID:SCR_005547]。任何刺激重复,如果 PSD 在任何频率点与平均值偏离 6 SD,则被认为是该电极的不良重复。




 


图 2. 为每个单极电极分离不良重复的流程图。


常见的错误重复集(图 3):接下来,我们在所有 64 个单极电极上创建了一组常见的错误重复,如下所示:
我们首先丢弃了那些超过 30% 的重复标记为坏的电极(来自步骤 a)。
然后,我们通过在用于分析的十个单极电极(P3、P1、P2、P4、PO3、POz、PO4 、O1、Oz 和 O2)。
接下来,如果重复出现在剩余电极总数的 10% 以上,我们会在该组中包含任何重复(在步骤 a 中标记为错误)。这是我们丢弃的最后一组错误重复,因此其余的刺激重复用于分析。


 


图 3. 在所有可分析的单极电极上创建一组常见的错误重复的流程图。


进一步拒绝电极(图 4):如上所述,我们丢弃了那些阻抗超过 25 KΩ 的电极,或者那些超过 30% 的重复标记为坏的电极。有时,我们观察到 PSD 趋于变平的电极(例如,如果它达到本底噪声,就会发生这种情况)。因此,我们添加了另一条管道来拒绝此类电极。为此,我们计算了每个单极电极的 PSD 斜率,如下所示:
我们使用一个锥度(TW=1)估计每个可分析重复的 PSD,并对得到的 PSD 进行平均。
为这些试验平均的 PSD拟合了一个幂律函数,其中P(f)=A.f^(-β)A (比例因子)和β (斜率)是使用最小二乘法最小化(使用 MATLAB 中的fminsearch函数)获得的自由参数, P是跨频率f∈[56 84]赫兹。
我们丢弃了那些斜率小于 0 的电极。我们对每个块分别执行此操作。


 


图 4. 丢弃不良单极电极的流程图。


对于任何块,如果没有可分析的双极电极(参见下面的数据分析小节),我们为该对象丢弃了该块,并将其余块中的数据汇集起来进行数据分析。对于Murty等人在拒绝块方面的细微差别。 (2020, 2021) ,请参阅各自的文章。




数据分析


电极的选择


我们将每个电极的离线数据重新参考到其相邻电极(称为双极参考的参考方案),如图 5 所示(Murty等人,2020) 。这样做是因为我们发现伽马振荡在双极参考方案中更为突出(参见 Murty等人的图 1,2020 )。具体来说,我们考虑了以下九种双极电极进行分析: PO3-P1、PO3-P3、POz-PO3、PO4-P2、PO4-P4、POz-PO4、Oz-POz、Oz-O1 和 Oz-O2 。如果双极电极的任何一个构成单极电极在伪影剔除期间被标记为坏,我们就丢弃了双极电极。


 


图 5. 显示双极蒙太奇的示意图。
根据国际 10-10 系统(Easycap GmbH,德国),彩色电极是放置在头皮上的物理电极。这些电极在线参考 FCz(单极参考方案)。地面位于 AFz。来自相邻物理电极的数据相互离线参考(双极参考方案)。双极电极以红色显示。每个双极电极实际上放置在两个组成的单极电极之间。我们在研究中分析了 112 个这样的电极。


脑电数据分析


我们使用 MATLAB 编写的自定义代码(The MathWorks, Inc.,RRID:SCR_001622)分析了所有数据,如 Murty等人所述。 (2020 年)。我们使用如下所述的 Chronux 工具箱,使用具有单个 DPSS 锥度 (TW=1) 的多锥度方法计算 PSD 和时频功率谱图。我们对所有可分析刺激重复的原始 PSD 和原始频谱图进行平均,以进行最终分析。此外,我们使用EEGLAB 工具箱 (Delorme and Makeig, 2004, RRID:SCR_007292)的topoplot.m函数生成头皮图,修改为将每个电极显示为彩色圆盘。
原始 PSD 的计算和 PSD 的变化: 我们分别计算了基线期(-500 ms 和 0 ms 的刺激开始)和刺激期(250 ms 和 750 ms)在每个频率下的功率(分辨率为 2 Hz),避免了 0 到 0 之间与刺激开始相关的瞬态250 毫秒)。对于 PSD 从基线到刺激期的变化(以 dB 表示),我们使用以下等式:


ΔP(f)=10(log_10  (P_st (f))/(P_bl (f)))


其中P(f)是频率f 处的功率。图 6a 显示了刺激(实线)和基线(虚线)周期的原始 PSD(黑线),以及示例对象的 PSD(蓝色实线)变化。在 20-34 Hz(慢伽马)和 36-66 Hz(快伽马)频率范围内观察到明显的伽马振荡作为“颠簸”。
计算原始时频功率谱图和变化:我们在刺激开始的 -500 ms 和 1,200 ms 期间估计了原始时频功率谱图。我们选择了 250 ms 的移动窗口和 25 ms 的步长,频率分辨率为 4 Hz。为了计算功率谱图的变化(以 dB 表示),我们从每个频率的整个原始谱图中减去原始谱图基线周期内的平均功率。换句话说,我们使用了以下公式:


ΔS(t,f)=10(log_10  S(t,f)/(S_bl (f) ))


其中S_bl (f)是频率 (f) 处的基线功率(从 -0.5 到 0 s 的平均值)。图 6b 显示了与图 6a 相同的示例主题的原始频谱图(顶行)和功率频谱图的变化(底行)。在功率谱图的变化中稳健地观察到慢速和快速伽马振荡。
计算不同频带的原始功率:为了估计基线和刺激期间感兴趣频带(即阿尔法、慢伽马和快伽马)的原始功率,我们将估计的各个频带中各个频率仓的功率相加在 PSD 中。换句话说,我们使用了以下等式:


Raw power= ∑_(f=a)^b▒〖P(f)〗


其中P(f)是频率f处的功率, a和b是感兴趣频带的下限和上限。
计算不同频带中刺激引起的功率变化:为了估计刺激期间与基线期相比的功率变化,我们对每个频带使用以下等式:


Change in power=10(log_10  〖Raw power〗_st/〖Raw power〗_bl )
图 6. 示例主题中的频谱和频谱图。
a) 显示刺激(黑色实线)和基线(黑色虚线)周期中的原始 PSD(左轴)的图,在九个双极电极(PO3-P1、PO3-P3、POz-PO3、PO4-P2、PO4 -P4、Poz-PO4、Oz-Poz、Oz-O1 和 Oz-O2),并以对数10表示。在右侧的轴上,PSD(蓝色实线)从刺激到基线周期的变化以 dB 标度显示。慢伽马 (20–34 Hz) 和快伽马 (36–66 Hz) 波段分别以垂直的紫色和橙色线显示。 b) 显示原始(上图)和功率时频谱图(下图)变化的图,与 a 中的数据相同。伽马波段显示为纯白色(慢伽马)和虚线(快速伽马)。垂直黑色虚线表示向受试者呈现刺激的时间(0-0.8 s),而红色垂直虚线表示考虑用于光谱分析的刺激期(0.25-0.75 s)。


这些分析的代码可在以下网址获得: https ://github.com/supratimray/TLSAEEGProjectPrograms 。
工件拒绝管道的代码: https ://github.com/supratimray/CommonPrograms在 ReadData/findBadTrialsEEG 下。




致谢


这项工作得到了塔塔信托基金、威康信托/DBT 印度联盟(中级奖学金 500145/Z/09/Z 和高级奖学金 IA/S/18/2/504003 到 SR)和 DBT-IISc 合作伙伴计划(到 SR)。与该协议相关的原始研究文章如下:
Murty, DVPS, Manikandan, K., Kumar, WS, Ramesh, RG, Purokayastha, S., Nagendra, B., Ml, A., Balakrishnan, A., Javali, M., Rao, NP 和 Ray, S. (2021 年)。在患有轻度认知障碍和阿尔茨海默病的老年人中,刺激诱导的伽马节律较弱。 生命 10:e61666 。




利益争夺


作者声明没有竞争的经济利益。




伦理


所有受试者自愿参加我们的研究,并为他们的时间和努力获得金钱补偿。我们在实验前获得了所有受试者的知情同意。印度科学研究所人类伦理委员会、NIMHANS 和班加罗尔 MS Ramaiah 医院批准了所有程序。




参考


Amir, A.、Headley, DB、Lee, SC、Haufler, D. 和 Pare, D. (2018)。警觉相关的伽马振荡协调基底外侧杏仁核神经元的整体活动。 神经元97(3):656-669 e657。
An, KM, Ikeda, T., Yoshimura, Y., Hasegawa, C., Saito, DN, Kumazaki, H., Hirosawa, T., Minabe, Y. 和 Kikuchi, M. (2018)。自闭症谱系障碍儿童运动控制过程中改变的伽马振荡。 神经科学杂志 38(36):7878-7886。
Buzsáki , G. 和 Wang, XJ (2012)。伽马振荡机制。 Annu Rev Neurosci 35:203-225。
Chalk, M., Herrero, JL, Gieselmann, MA, Delicato, LS, Gotthardt, S. 和 Thiele, A. (2010)。注意减少了刺激驱动的伽马频率振荡和 V1 中的尖峰场相干性。 神经元66(1):114-125。
Colgin, LL, Denninger, T., Fyhn, M., Hafting, T., Bonnevie, T., Jensen, O., Moser, MB 和 Moser, EI (2009)。伽马振荡的频率会引导海马体中的信息流。 自然462(7271):353-357。
Gieselmann, MA 和 Thiele, A. (2008)。猕猴 V1 尖峰活动的空间整合和环绕抑制特性与局部场电位的比较。 Eur J Neurosci 28(3):447-459。
Gray, CM, Konig, P., Engel, AK 和 Singer, W. (1989)。猫视觉皮层的振荡反应表现出反映全局刺激特性的列间同步。 自然338(6213):334-337。
Gregoriou, GG, Gotts, SJ, Zhou, H. 和 Desimone, R. (2009)。注意期间前额叶和视觉皮层之间的高频、长程耦合。 科学324(5931):1207-1210。
Hirano, Y.、Oribe, N.、Kanba, S.、Onitsuka, T.、Nestor, PG 和 Spencer, KM (2015)。精神分裂症的自发伽马活动。 JAMA 精神病学72(8):813-821。
Iaccarino,HF,Singer,AC,Martorell,AJ,Rudenko,A.,Gao,F.,Gillingham,TZ,Mathys,H.,Seo,J.,Kritskiy,O.,Abdurrob,F.,等。 (2016 年)。伽马频率夹带减弱淀粉样蛋白负荷并改变小胶质细胞。 自然540(7632):230-235。
Juergens, E.、Guettler, A. 和 Eckhorn, R. (1999)。视觉刺激在猴子皮质内和脑电图电位中引发锁定和诱导的伽马振荡,但在人类脑电图中没有。 Exp Brain Res 129(2):247-259。
凯,LM(2003)。嗅球中的两种伽马振荡:对行为状态和突触相互作用的依赖。 J Integr Neurosci 2(1):31-44。
Koch, SP、Werner, P.、Steinbrink, J.、Fries, P. 和 Obrig, H. (2009)。刺激诱导和状态依赖的持续伽马活动与人类的血流动力学反应紧密相关。 神经科学杂志 29(44):13962-13970。
Kumar, WS, Manikandan, K., Murty, DVPS, Ramesh, RG, Purokayastha, S., Javali, M., Rao, NP 和 Ray, S. (2021)。刺激引起的窄带伽马振荡在人类脑电图中的健康老年人中是可靠的。 bioRxiv :2021.2007.2006.451226。
Mably, AJ 和 Colgin, LL (2018)。认知障碍中的伽马振荡。 Curr Opin Neurobiol 52:182-187。
Mitra, P. 和 Bokil, H. (2008)。观察大脑动态。 牛津,纽约:牛津大学出版社。
Murty, DVPS, Manikandan, K., Kumar, WS, Ramesh, RG, Purokayastha, S., Nagendra, B., Ml, A., Balakrishnan, A., Javali, M., Rao, NP 和 Ray, S. (2021 年)。在患有轻度认知障碍和阿尔茨海默病的老年人中,刺激诱导的伽马节律较弱。 生命 10:e61666 。
Murty, DVPS, Manikandan, K., Kumar, WS, Ramesh, RG, Purokayastha, S., Javali, M., Rao, NP 和 Ray, S. (2020)。人类脑电图中健康老年人的伽马振荡随着年龄的增长而减弱。 神经影像215:116826。
Murty, DVPS, Shirhatti, V., Ravishankar, P. 和 Ray, S. (2018)。大型视觉刺激会在灵长类动物视觉皮层中引发两种不同的伽马振荡。 神经科学杂志 38(11):2730-2744。
Muthukumaraswamy,SD 和辛格,KD(2013 年)。视觉伽马振荡:刺激类型、视野覆盖和刺激运动对 MEG 和 EEG 记录的影响。 神经影像69:223-230。
Orekhova, EV, Butorina, AV, Sysoeva, OV, Prokofyev, AO, Nikolaeva, AY 和 Stroganova, TA (2015)。人类的伽马振荡频率受视觉运动速度的调节。 神经生理学杂志114(1):244-255。
Palop, JJ 和 Mucke, L. (2016)。阿尔茨海默病中的网络异常和中间神经元功能障碍。 Nat Rev Neurosci 17(12):777-792。
Pattisapu, S. 和 Ray, S. (2021)。可以使用开放硬件低成本脑电图放大器记录刺激引起的人体窄带伽马振荡。 doi:https://doi.org/10.1101/2021.11.16.468841。
Pesaran, B., Pezaris, JS, Sahani, M., Mitra, PP 和 Andersen, RA (2002)。猕猴顶叶皮层工作记忆中神经元活动的时间结构。 Nat Neurosci 5(8):805-811。
Ray, S. 和 Maunsell, JH (2010)。视觉皮层伽马频率的差异限制了它们在计算中的可能用途。 神经元67(5):885-896。
Ray, S. 和 Maunsell, JH (2011)。猕猴视觉皮层伽马节律和高伽马活动的不同起源。 公共科学图书馆生物学9(4):e1000610。
Shirhatti, V. 和 Ray, S. (2018)。长波长(微红色)色调会在灵长类动物初级视觉皮层中引起异常大的伽马振荡。 Proc Natl Acad Sci USA 115(17): 4489-4494。
Siegel, M. 和König , P. (2003)。由清醒行为猫的区域 18 中的刺激依赖性同步定义的功能性伽马带。 神经科学杂志 23(10):4251-4260。
Tada, M.、Nagai, T.、Kirihara, K.、Koike, S.、Suga, M.、Araki, T.、Kobayashi, T. 和 Kasai, K.(2016 年)。发病前高危个体与首发精神分裂症之间听觉伽马振荡反应的差异变化。 大脑皮层26(3):1027-1035。
Uhlhaas, PJ 和 Singer, W. (2007)。神经同步障碍告诉我们关于自闭症的哪些信息? 生物学精神病学62(3):190-191。
van Pelt, S. 和 Fries, P. (2013)。视觉刺激偏心率影响人类伽马峰值频率。 神经影像 78:439-447。
Veit, J.、Hakim, R.、Jadi, MP、Sejnowski, TJ 和 Adesnik, H. (2017)。通过生长抑素中间神经元实现皮质伽马带同步。 自然神经科学20(7):951-959。
Wilson, TW, Rojas, DC, Reite, ML, Teale, PD 和 Rogers, SJ (2007)。自闭症儿童和青少年表现出降低的 MEG 稳态伽马反应。 生物学精神病学62(3):192-197。

  • English
  • 中文翻译
免责声明 × 为了向广大用户提供经翻译的内容,www.bio-protocol.org 采用人工翻译与计算机翻译结合的技术翻译了本文章。基于计算机的翻译质量再高,也不及 100% 的人工翻译的质量。为此,我们始终建议用户参考原始英文版本。 Bio-protocol., LLC对翻译版本的准确性不承担任何责任。
Copyright Murty and Ray. This article is distributed under the terms of the Creative Commons Attribution License (CC BY 4.0).
引用: Readers should cite both the Bio-protocol article and the original research article where this protocol was used:
  1. Murty, D. V. P. S. and Ray, S. (2022). Stimulus-induced Robust Narrow-band Gamma Oscillations in Human EEG Using Cartesian Gratings. Bio-protocol 12(7): e4379. DOI: 10.21769/BioProtoc.4379.
  2. Murty, D. V. P. S., Manikandan, K., Kumar, W. S., Ramesh, R. G., Purokayastha, S., Nagendra, B., Ml, A., Balakrishnan, A., Javali, M., Rao, N. P. and Ray, S. (2021). Stimulus-induced gamma rhythms are weaker in human elderly with mild cognitive impairment and Alzheimer's disease. Elife 10: e61666.
提问与回复

如果您对本实验方案有任何疑问/意见, 强烈建议您发布在此处。我们将邀请本文作者以及部分用户回答您的问题/意见。为了作者与用户间沟通流畅(作者能准确理解您所遇到的问题并给与正确的建议),我们鼓励用户用图片的形式来说明遇到的问题。

如果您对本实验方案有任何疑问/意见, 强烈建议您发布在此处。我们将邀请本文作者以及部分用户回答您的问题/意见。为了作者与用户间沟通流畅(作者能准确理解您所遇到的问题并给与正确的建议),我们鼓励用户用图片的形式来说明遇到的问题。