Stimulus-specific Cortical Visual Evoked Potential Morphological Patterns

Summary

In this paper, we present a protocol to investigate differential cortical visual evoked potential morphological patterns through stimulation of ventral and dorsal networks using high-density EEG. Visual object and motion stimulus paradigms, with and without temporal jitter, are described. Visual evoked potential morphological analyses are also outlined.  

Abstract

This paper presents a methodology for the recording and analysis of cortical visual evoked potentials (CVEPs) in response to various visual stimuli using 128-channel high-density electroencephalography (EEG). The specific aim of the described stimuli and analyses is to examine whether it is feasible to replicate previously reported CVEP morphological patterns elicited by an apparent motion stimulus, designed to simultaneously stimulate both ventral and dorsal central visual networks, using object and motion stimuli designed to separately stimulate ventral and dorsal visual cortical networks.  Four visual paradigms are presented: 1. Randomized visual objects with consistent temporal presentation. 2. Randomized visual objects with inconsistent temporal presentation (or jitter).  3. Visual motion via a radial field of coherent central dot motion without jitter.  4. Visual motion via a radial field of coherent central dot motion with jitter.  These four paradigms are presented in a pseudo-randomized order for each participant.  Jitter is introduced in order to view how possible anticipatory-related effects may affect the morphology of the object-onset and motion-onset CVEP response.  EEG data analyses are described in detail, including steps of data exportation from and importation to signal processing platforms, bad channel identification and removal, artifact rejection, averaging, and categorization of average CVEP morphological pattern type based upon latency ranges of component peaks. Representative data show that the methodological approach is indeed sensitive in eliciting differential object-onset and motion-onset CVEP morphological patterns and may, therefore, be useful in addressing the larger research aim. Given the high temporal resolution of EEG and the possible application of high-density EEG in source localization analyses, this protocol is ideal for the investigation of distinct CVEP morphological patterns and the underlying neural mechanisms which generate these differential responses.      

Introduction

Electroencephalography (EEG) is a tool that offers an inexpensive and non-invasive approach to the study of cortical processing, especially when compared to cortical assessment methods such as functional magnetic resonance imaging (fMRI), positron emission tomography (PET), and diffusion tensor imaging (DTI)¹. EEG also provides high temporal resolution, which is not possible to attain when using measures such as fMRI, PET, or DTI². High temporal resolution is critical when examining central temporal function in order to obtain millisecond-precision of neurophysiologic mechanisms related to the processing of specific input or events.  In the central visual system, cortical visual evoked potentials (CVEPs) are a popular approach in studying time-locked neural processes in the cerebral cortex.  CVEP responses are recorded and averaged over a number of event trials, resulting in peak components (e.g., P1, N1, P2) arising at specific millisecond intervals. The timing and amplitude of these peak neural responses can provide information concerning cortical processing speed and maturation, as well as deficits in cortical function^3,4,5.  

CVEPs are specific to the type of visual input presented to the viewer. Using certain stimuli in a CVEP paradigm, it is possible to observe the function of distinct visual networks such as the ventral stream, involved in processing form and color, or parvocellular and magnocellular input^6,7,8, and the dorsal stream, which largely processes motion or magnocellular input^9,10. CVEPs generated by these networks have been useful not only in better understanding typical neurophysiologic mechanisms underlying behavior but also in the targeted treatment of atypical behaviors in clinical populations. For example, delayed CVEP components in both dorsal and ventral networks have been reported in children with dyslexia, which suggests that visual function in both these networks should be targeted when designing an intervention plan¹¹.  Thus, CVEPs recorded via EEG offer a powerful clinical tool through which to assess both typical and atypical visual processes. 

In a recent study, high-density EEG was used to measure the apparent motion-onset CVEPs in typically developing children, with the goal of examining variable CVEP responses and related visual cortical generators across development. Participants passively viewed apparent motion stimuli^12,13,14,15, which consisted of both shape change and motion, designed to simultaneously stimulate dorsal and ventral streams. It was found that approximately half of the children responded with a CVEP waveform shape, or morphology, consisting of three peaks (P1-N1-P2, pattern A).  This morphology is a classic CVEP response observed throughout the literature. In contrast, the other half of the children presented with a morphological pattern comprised of five peaks (P1-N1a-P2a-N1b-P2b, pattern B). To our knowledge, the robust occurrence and comparison of these morphological patterns have not previously been discussed in CVEP literature in either child or adult populations, although variable morphology has been noted in both apparent-motion and motion-onset CVEPs^14,16. Furthermore, these morphological differences would not have been apparent in research using other cortical functional assessment methods, such as fMRI or PET, due to the low temporal resolution of these measures.

To determine the cortical generators of each peak in CVEP patterns A and B, source localization analyses were performed, which is a statistical approach used to estimate the most likely cortical regions involved in the CVEP response^12,13. For each peak, regardless of the morphological pattern, primary and higher-order visual cortices were identified as sources of the CVEP signal.  Thus, it appears that the main difference underlying CVEP morphology elicited by apparent motion is that those with pattern B activate visual cortical regions additional times during processing. Because these types of patterns have not been previously identified in the literature, the purpose of the additional visual processing in those with CVEP pattern B remains unclear.  Therefore, the next aim in this line of research is to gain a better understanding of the cause of the differential CVEP morphology and whether such patterns may relate to visual behavior in both typical and clinical populations. 

The first step in understanding why some individuals might demonstrate one CVEP morphology versus another is to determine whether these responses are intrinsic or extrinsic in nature.  In other words, if an individual demonstrates one pattern in response to a visual stimulus, will they respond with a similar pattern to all stimuli?  Or is this response stimulus-dependent, specific to the visual network or networks activated?

To answer this question, two passive visual paradigms were designed, intended to separately activate specific visual networks. The stimulus presented in the initial study was designed to stimulate both dorsal and ventral streams simultaneously; thus, it was unknown if one or both networks were involved in generating specific waveform morphology. In the current methodological approach, the paradigm designed to stimulate the ventral stream is composed of highly identifiable objects in basic shapes of squares and circles, eliciting object-onset CVEPs. The paradigm designed to stimulate the dorsal stream consists of visual motion via a radial field of coherent central dot motion dots at a fixed speed toward a fixation point, eliciting motion-onset CVEPs. 

A second question that arose as a result of the initial study was whether differential VEP morphology could be due to participant anticipation of upcoming stimuli¹³. For instance, research has shown that top-down cortical oscillatory activity occurring prior to a target stimulus may predict subsequent CVEP and behavioral responses to some degree^17,18,19. The apparent motion paradigm in the first study employed non-randomized frames of a radial star and circle with consistent inter-stimulus intervals (ISIs) of 600 ms. This design may have encouraged the expectation and prediction of the upcoming stimulus, with resulting oscillatory activity affecting subsequent CVEP morphology^12,13,19.  

To address this issue, the visual object and motion paradigms in the current protocol are designed with both consistent ISIs of the same temporal value and randomized ISIs with different temporal values (i.e., jitter).  Using this approach, it may be possible to determine how temporal variation can affect VEP morphology within distinct visual networks. Altogether, the aim of the described protocol is to determine if the visual object and motion stimuli would be sensitive to variations in CVEP morphology and whether the temporal variation of stimuli presentation would affect characteristics of the CVEP response, including peak latency, amplitude, and morphology. For the purpose of the current paper, the goal is to determine the feasibility of the methodological approach. It is hypothesized that both visual objects and motion may elicit variable morphology (i.e., patterns A and B will be observed across subjects in response to both stimuli) and that temporal variation would affect object-onset and motion-onset CVEP components. 

Protocol

All methods described here have been approved by the Institutional Review Board (IRB) for Human Research at the University of Texas at Austin.

1. Stimuli Characteristics

1. Create object stimuli using open source images available through the Bank of Standardized Stimuli (BOSS). This database consists of standardized images used throughout visual cognitive experiments.  Download four images (e.g., ball02, book01a, brick, button03) with a high rate of identification (above 75%)^20,21.  
2. Create motion stimuli using a modified version of the DotDemo script, which is available through the open source Psychtoolbox-3 set of functions operated via MATLAB, as well as the movie function available in MATLAB (see the Supplementary File).
   1. Configure the dot field parameters according to the size of the presentation screen and viewing distance.  
   2. Enter 3600 for the number of movie frames. 
   3. Enter 80 (in cm) for monitor width.
   4. Enter dot speed at 5°/s.
   5. Enter a dot limited lifetime fraction of 0.05.
   6. Enter 200 for the number of dots.
   7. Enter the minimum radius of the field annulus as 1° and the maximum as 15°.  
   8. Enter 0.2° for the width of each dot.
   9. Enter 0.35° for the radius of the fixation point.
   10. Specify that white dots are used on a black background.
   11. Export the movie in .avi format.

2. Visual Paradigm Design

1. Create paradigms via stimulus-presentation software. Generate fixation crosses with Courier New size 18 font, bold, and centered on the presentation screen.
2. Design the visual object paradigm without temporal jitter (i.e., consistent ISI values) by creating a black fixation cross on a white background presented for 500 ms, followed by one of four objects presented in randomized order: ball, book, brick, or button.  
   1. Present each object for 600 ms (Figure 1A).  Show all objects 75 times, for a total of 300 trials and a paradigm duration of 5.5 min.  
3. Design the visual object paradigm with temporal jitter to consist of the same black fixation cross on a white background, shown for a period of 500 or 1,000 ms and followed by one of the four objects, lasting for 600 or 1000 ms (Figure 1B).  
   1. Create four trials using stimulus-presentation software: a fixation cross with a duration of 500 ms, followed by an object for 600 ms; a fixation cross with a duration of 500 ms, followed by an object for 1,000 ms; a fixation cross with a duration of 1,000 ms, followed by an object for 600 ms; and a fixation cross with a duration of 1,000 ms followed by an object for 1,000 ms.  
      1. Randomize these trials. Present each trial 19 times, culminating in 304 trials and resulting in a viewing time of approximately 7.85 minutes.
4. Create the visual motion paradigm without temporal jitter by generating a white fixation cross centered on a black background, lasting for 500 ms, followed by the visual motion movie, which is truncated to present for approximately 1,000 ms (Figure 2A). 
   1. Repeat this sequence a total of 300 times, for a viewing duration of approximately 7.5 min.  
5. Create the visual motion paradigm with temporal jitter using the same fixation cross, lasting for intervals of 500, 750, or 1,000 ms.  
   1. After each fixation cross, present the visual motion movie with a duration of approximately 600 or 1,000 ms (Figure 2B).  
   2. Create six trials: A fixation cross with a duration of 500 ms, followed by a movie for 600 ms, a fixation cross with a duration of 750 ms, followed by a movie for 600 ms, a fixation cross with a duration of 1,000 ms, followed by a movie for 600 ms, a fixation cross with a duration of 500 ms followed by a movie for 1000 ms, a fixation cross with a duration of 750 ms followed by a movie for 1,000 ms and a fixation cross with a duration of 1,000 ms followed by a movie for 1,000 ms.  
      1. Randomize these trials, with each shown 50 times.  Present a total of 300 trials, for a viewing period of approximately 7.75 min.

3. Participant Consent, Case History, and Vision Screening

1. Greet the participant on arrival. Obtain informed consent by presenting the participant with a consent for participation in research form. Explain the consent form to the participant and answer any questions that arise.
2. Have the participant fill out a case history form that includes information on native language, handedness, hearing status, vision status, and other diagnoses the participant may have (e.g., psychological and neurological). Exclude participants who report hearing loss and/or neurological diagnoses, such as traumatic brain injury.  Include all other participants.
3. Escort the participant out of the lab to complete a vision screening using a Snellen chart to determine visual acuity. Have the participant stand 20 feet away from the chart and begin by covering his or her left eye to determine right eye visual acuity, and then switch eyes to determine left eye visual acuity. Calculate visual acuity based on the smallest line of text that the participant can repeat at least one more than half of the total number of letters. 
   NOTE: For example, if the participant repeats 5 of the 8 letters on the 20/20 line, visual acuity is calculated as 20/20 in that eye.
4. Escort the participant into the EEG recording room. Have the participant sit in the designated chair in the center of a double-walled magnetic-shielded soundproof booth.  

4. EEG Preparation

1. Measure the head circumference of the participant in centimeters and select the appropriate EEG net size. Measure and mark the midpoint of the scalp (midway between nasion/inion and right and left mastoids) for the placement of the reference electrode.
2. Prepare a solution of warm water (1 L) mixed with baby shampoo (5 mL) and potassium chloride (11 g/10 cc), which increases the electrical conductance between the electrodes and scalp, leading to lower voltage impedances and an increased signal-to-noise ratio.
3. Place the EEG net in the solution. Allow the net to soak in the solution for 5 min before placing on the participant’s scalp.
4. Turn on the stimulus-presentation computer and the EEG acquisition computer.
5. Place a towel or other absorbent material around the participant’s neck to prevent the solution from dripping onto his or her clothes. 
6. Connect the EEG net to the amplifier. Instruct the participant to close his or her eyes when putting on the EEG net to prevent the solution from dripping into his or her eyes.
7. Firmly grip the EEG net with both hands and spread into place onto the participant’s head. Ensure that the net is placed symmetrically on the scalp head, with the reference electrode at the scalp midline point that was measured. Tighten the chin and ocular net lines to ensure a secure connection between the scalp and electrodes. Ask the participant if he or she is comfortable and if anything needs to be adjusted.
8. Check for the proper electrode impedance values, with an average target of 10 kΩ.
9. To reduce impedance values following the placement of the electrode net, use a 1 mL pipette to apply the potassium chloride solution onto the scalp/electrodes that have a high impedance. Continue this process until adequate impedances values across the electrodes are achieved. 

5. EEG Recording

1. Instruct the participant to focus on the visual stimuli that will appear on the monitor. The viewing distance is approximately 65 inches.
2. Use a pseudorandom number generator to determine the order of presentation for the four visual paradigms. 
3. Begin the visual tasks and EEG recording.
4. Monitor the EEG recording as necessary. If ongoing EEG shows high myogenic or 60 Hz activity, pause the experiment to recheck electrode-scalp connectivity.
5. Repeat steps 5.3 and 5.4 for the visual object paradigm, the visual object with the temporal jitter paradigm, the visual motion paradigm, and the visual motion with temporal jitter paradigm.
6. At the conclusion of the experiment, instruct the participant to close his or her eyes in order to prevent the solution from entering his or her eyes when removing the net. Begin by loosening the chin and ocular net lines, then remove the net by gently pulling the chin strap up and over the participant’s head, making sure to pull slowly to ensure the net will not get tangled in the participant’s hair. 
7. Disconnect the EEG net from the amplifier. Begin the disinfection process by placing the EEG cap in and out of a bucket filled with water and rinsing under a faucet. Then, create the disinfectant solution by adding approximately 2 l of water to the disinfectant bucket and mixing 15 ml of disinfectant with the water. 
8. Immerse the sensor end of the net in the disinfectant. Set a timer for 10 min; for the first 2 min, continuously plunge the net up and down. Leave the net soaking for the remainder of the 10 min.
9. Remove EEG net from disinfectant solution. Place the EEG net in and out of the electrode bucket filled with water and under running water to rinse. Repeat four times.  Allow the net to air dry. 

6. EEG Analyses

1. Export EEG files for analyses in MATLAB via the EEGLAB toolbox using a 1 Hz high-pass filter, segmentation around each trial (or event) of 100 ms pre-stimulus and 500 ms post-stimulus periods.  
2. Import data using the EEGLAB toolbox.  
   1. Choose the File option from the drop-down menu and click on Import data.  Select using EEGLAB Functions and plugins from the menu.  Next click on the appropriate export file format.
3. Re-assign channel locations based on the type of electrode montage used by choosing Edit from the drop-down menu and selecting Channel locations.  Click on Look up locs and select the ellipses to locate the path of the electrode montage file of interest. 
4. Assign pre- and post-stimulus times to the epoch start and end times. Enter a value of -0.1 s in the Start time box. 
5. Baseline-correct data according to the pre-stimulus interval.
6. Identify and remove bad channels using probability at a Z-score threshold of 2.5. 
   1. Verify successful identification and removal of bad channels by plotting all electrodes. Manually remove channels with mean voltage amplitudes outside of the range of +/- 30 µV.
7. Perform artifact rejection by entering values of -100 µV and +100 µV.
   NOTE: This method is effective in the removal of ocular activity recorded at ocular electrodes (126, 127). However, it may be necessary to manually remove trials with artifact occurring at small-voltage amplitude (i.e., within the +/- 100 µV range) for certain participants. 
   1. Take note of channels that were bad for entire segments (i.e., with voltages outside of the +/-100 µV range) and highlighted in red. Manually remove these bad channels if they constitute 60% or more of the rejected trials. Repeat this step as many times as necessary. 
   2. Follow artifact removal steps as previously described. Ensure that a minimum of 100 sweeps are accepted. Remove trials marked for rejection.
8. Plot channel 75 (equivalent to Oz), or the channel(s) of interest, to categorize morphological patterns. Prior to plotting this channel, make sure to perform pre-stimulus baseline correction.  
9. Choose pattern A if CVEP morphology is characterized by a large positive peak at approximately 100-115 ms (P1), followed by a negative peak at approximately 140-180 ms (N1) and a positive peak at approximately 165-240 ms (P2).  
10. Choose pattern B if CVEP morphology is characterized by a large positive peak at approximately 100-115 ms (P1), followed by a negative peak at approximately 140-180 ms (N1a), a positive peak at approximately 180-240 ms (P2a), then a negative peak at approximately 230-280 ms (N1b) and positive peak at approximately 260-350 ms (P2b).    
11. Append individual datasets together according to the morphological pattern visually observed to create a group average. Name and save the newly merged dataset file.  
12. View appended files as an average by plotting the channel(s) of interest.

Results

Figure 3 and Figure 4 show the representative object-onset and motion-onset CVEP results of five participants, aged 19-24 years, who passively viewed each visual paradigm. This design allowed observation of CVEP responses elicited by visual objects (with and without jitter) and visual motion (with and without jitter) both within and across subjects according to each condition.  Participant CVEPs were grouped according to the morphological pattern elicited b...

Discussion

The goal of this methodological report was to evaluate the feasibility in recording differential CVEP morphology by using visual object and motion stimuli specifically designed to separately stimulate ventral and dorsal streams in passive viewing tasks^6,7,8, both with and without variation of ISIs (jitter)¹⁹. Conditions were not designed to be directly compared, rather, observations were made as to whethe...

Materials

Name	Company	Catalog Number	Comments
E-Prime 2.0	Psychology Software Tools, Inc		Used in data acquisition
Net Amps 400	Electrical Geodesics, Inc		Used in data acquisition
Net Station Acquisition V5.2.0.2	Electrical Geodesics, Inc		Used in data acqusition
iMac (27-inch)	Apple		Used in data acquisition
Optiplex 7020 Computer	Dell		Stimulus computer
HydroCel GSN EEG net	Electrical Geodesics, Inc		Used in data acqusition
1 ml pipette	Electrical Geodesics, Inc		Used to lower impedances
Johnson's Baby Shampoo	Johnson & Johnson		Used in impedance solution
Potassium Chloride (dry)	Electrical Geodesics, Inc		Used in impedance solution
Control III Disinfectant Germicide	Control III		Used in disinfectant solution
32-inch LCD monitor	Vizio		Used to present stimuli
Matlab (R2016b)	MathWorks		Used in data analysis
EEGlab v14.1.2	Swartz Center for Computational Neuroscience, University of California, San Diego	https://sccn.ucsd.edu/eeglab/index.php	Used in data analysis
BOSS Database	Bank of Standardized Stimuli	https://sites.google.com/site/bosstimuli/	Used in generation of visual object stimuli
Psychtoolbox-3	Psychophysics Toolbox Version 3 (PTB-3)	http://psychtoolbox.org/	Used in generation of visual motion stimuli