Skip to content

21. Sample data: Steady state visual potential

Jump to bottom
Fa-Hsuan Lin edited this page Apr 4, 2024 · 14 revisions

1. Introduction

These are part of the data for simultaneous EEG-MRI acquisition. The experiment design was presenting checkerboard flashing (7.5 Hz reversal rate; 1000 ms duration) randomly to elicit steady-state evoked potentials (SSVEPs). Detailed of this experiment was described [here]. Data can be accessed by [University of Toronto OneDrive folder].

2. Data structure

EEG and MRI from three participants (coded by 180322_PYW, 180324_JLJ, and 180330_SYH) are included in this data sets. These data are stored in the following structure:

  • 180322_PYW -- analysis -- digitizer -- eeg_raw -- fmri_analysis -- fmri_raw

  • 180324_JLJ -- analysis -- digitizer -- eeg_raw -- fmri_analysis -- fmri_raw

  • 180330_SYH -- analysis -- digitizer -- eeg_raw -- fmri_analysis -- fmri_raw

  • subjects -- 180322_PYW -- 180324_JLJ -- 180330_SYH

The "subjects" folder includes the FreeSurfer reconstruction of anatomical brain MRI scans for each participants. Inside each subject's folder, there are sub-folders. Their names and contents are:

  • "analysis": EEG source analysis script and results
  • "eeg_raw": EEG measurements
  • "digitzer": EEG electrode locations
  • "fmri_analysis": fMRI analysis scripts and results
  • "fmri_raw": fMRI data

We expect to observe the neural and hemodynamic activity elicited by to the visual stimulation by EEG and fMRI, respectively.

3. EEG data

Evoked responses over the scalp

These are EEG data collected outside MRI. Thus they had no artifacts caused by the MRI scanner. Procedures to estimate neural current distributions of the EEG data are:

  1. Estimate the evoked response: [read_erp_outside.m].
  2. Render the evoked response topology and time course: [show_erp_scalp.m]. This script needs a [Matlab data file] for the definition of EEG electrode locations and a head model.

EEG source analysis

NOTE The following are procedures for EEG source analysis. Please refer [here] for some details. Please refer to [the procedure to create brain models from MRI]. Please note that some details may be different.

Structural MRI data for this exemplary subject can be downloaded [here].

  1. Co-register between EEG electrodes and the head by [this script].
  1. Prepare EEG source space (likely brain areas generating EEG signals) and brain model by [this script]

  1. Calculate the EEG forward solution. Sequentially [calculate the forward solution by OpenMEEG package] and [collect the calculated results].

  2. Calculate the EEG source by Minimum norm estimate using [this script].

  3. Render source estimates using [this script].

Alternative approaches are described below:

3b. Prepare the EEG forward solution by co-registering EEG sensor locations with the head model: [register_eeg_041018.m].

4b + 5b. Calculate the EEG forward solution: [make_fwd_041018.m].

NOTE This step requires the access to the Helsinki boundaryelement method (BEM) framework for MEG/EEG ([Stenroos et al., 2007], [Stenroos and Nummenmaa, 2016]). You can access the matlab toolbox inside Sunnybrook at ~fhlin/matlab/toolbox/distribution_pcore. If you don't have the access, just use the pre-calculated results [fwd_041018.mat].

6b. Estimate the neural current distributions: [make_erp_mne_051220.m].

7b. Upon completing these procedures, results can be visualized at Matlab: etc_render_fsbrain('subject','180322_PYW');

Left: The medial aspect of the inflated left hemisphere cortex for subject 180322_PYW overlaid with the y-directional component of the estimated neural current distributions for SSVEP at about 100 ms after the stimulus onset. Right: The time course of the y-directional component of the estimated neural current distributions for SSVEP at the left visual cortex.

4. fMRI data

Procedures to estimate the activated brain area from the fMRI data are:

  1. Unpack fMRI images from DICOM format to our preferred format. This is done by using the unpack script in FS-FAST. Check [this page] for detais. Here is the [unpack rule file] used for subject 180322_PYW.
  2. Pre-processing of data, including motion correction, slice-timing correction, and spatial normalization. Please check [the page].
  3. Register between fMRI and structural MRI. Three registration should be done for each subject, because there were three runs of data for each subject. Here is the [example script] for run 1 for subject 180322_PYW.
  4. Morph (Spatially normalize) the fMRI data to an atlas. Here we used the subject fsaverage in FreeSufer to transform the fMRI data into a standard atlas space (MNI305). We used a [Matlab script] to do this job.
  5. Because this is an event-related fMRI design, we need to get onset timing of the visual stimuli, which were a brief episode of 8-Hz checkerboard flashing for 1 second. We used [this script] to prepare the stimuli onset timing. Here the event (trigger) coded in numbers 1 and 10 represented the showing of visual stimuli.
  6. Use General linear model to estimate brain areas activated by the stimuli. Please check [the page]. We used [this script];
  7. Examine the estimated activated brain area. We used the tool at [our toolbox]. Results were rendered by using [this script].

4. Simultaneous EEG-MRI data

We used temporally sparse simultaneous multi-slice inverse imaging (SMS-InI) to acquire the fMRI data (one volume every two-seconds) during concurrent EEG recording.

fMRI analysis : convert the "raw" data to values over brain surfaces (and volumes).

  1. [smsini_meas2mat.m] converts meas.dat files (from MRI TWIX) into matlab.

  2. [smsini_make_ref.m]. This script creates an image of the reference scan

  3. [register.script]. This script co-registers between the reference scan image and the FreeSurfer recon brain. Three runs of data will be registered separately.

NOTE Double check the registration with FreeView.

  1. [smsini_make_acc_surf.m]. This script converts the accelerated scan into dynamic values at cortical surfaces.

fMRI analysis : General Linear Modeling

  1. Read the stimulus onsets from EEG data using this script. This will create three .para files with stimulus onsets.

  2. Perform GLM using stimulus onsets using this script.

Clone this wiki locally