Epilepsy project: NYUAD collaboration with Cleveland Clinic Abu Dhabi

Overview

MEG protocol for participants with epilepsy on the KIT system. One session records an auditory task, a short artifact block, three visual/motor task runs, and a 60 min sleep recording, with marker-coil (mrk) measurements for head position between blocks.

All four tasks are existing lab experiments; this page gives the order, the settings to use for this project and the session checklist. Details of each task (stimuli, triggers, saved files) are on its experiment page.

Session protocol

#

Block

Experiment page

Default length

Recording

mrk-1

Marker-coil measurement

Before the first block

1

Auditory: two tones

Auditory two tones

2 min 50 s

Run 1, task twotones

2

Artifacts: tongue / swallow, eye movements, rapid eye blinks

No stimulus (see Artifact block)

about 2 min

Task artifacts

mrk-2

Marker-coil measurement

3

Motor: cued left/right button presses

Cued left/right button press

4 min 42 s

Run 1, task buttonlr

4

Visual categories: houses vs faces

Houses vs faces localizer

2 min 42 s

Run 1, task housesvsfaces

5

Early visual: checkerboard pattern reversal

Flipped squares

2 min 07 s

Run 1, task flippedsquares

mrk-3, mrk-4

Marker-coil measurements

Before the sleep block

6

Sleep, 60 min in four 15 min recordings

No stimulus (see Sleep recording)

4 x 15 min

Task sleep, runs 1-4, one mrk per 15 min

The task blocks take about 15 min of recording in total, plus the pauses between runs (each run is started separately). With the 60 min sleep recording, plan for about 1 h 30 min in the shielded room, plus participant preparation. The first session ran from mrk-1 to the start of sleep in 27 min (see Session log).

Note

Record each block as its own MEG file. The tasks reuse the same KIT trigger channels (224-229) for different events, so a trigger value only has a meaning together with the events.tsv of that run.

Head position (marker coils)

Measure the marker coils (mrk) at these points:

  • mrk-1 before the first block;

  • mrk-2 after the artifact block, before the motor task;

  • mrk-3 and mrk-4 after the checkerboard, before sleep;

  • one mrk every 15 min during sleep, i.e. one per sleep recording.

For co-registration, use the mrk measured closest to each recording. If the participant moved during a recording, use the mrk measured after the movement (see the note on sleep_4 in the Session log).

Photosensitivity

Every visual stimulus in this protocol is used with its photosensitivity-safe settings: constant mean luminance, low contrast, small central stimuli, gentle fades and at most 2 luminance changes per second. Text screens fade in and out, the screen fades to black before the window closes, and the Psychtoolbox start-up screens are disabled.

Important

  • In Flipped squares, keep “Participant has (or may have) photosensitive epilepsy” ticked (the default). The checkerboard is then a small central square (10 deg, about 5 % of the screen). Never tick “Full-screen checkerboard” in this project.

  • Keep the default contrast and rates in every task; the launchers refuse values outside their safety limits.

  • Set the stimulus PC desktop background on the projector screen to black, so opening or closing an experiment window is never a bright step.

  • Obtain the responsible clinician’s approval of the stimuli (Preview cues, Preview stimuli and Play test tones in the launchers) before the first participant, and keep a clinician or trained staff member available during the session.

  • ESC or Q stops any task immediately; the stimulus is removed and the partial data are saved.

Before the session

  1. Follow the operational protocol and the session checklist.

  2. On the stimulus PC, pull the latest code and run the self-test of each task once (no Psychtoolbox or DataPixx needed):

    cd experiments/psychtoolbox/auditory_two_tones;  main('validate')
cd ../button_press_lr;                           main('validate')
cd ../houses_vs_faces;                           main('validate')
cd ../flipped_squares;                           main('validate')

  3. Check that KIT trigger channels 224 to 229 all respond, with test_all_meg_channels_triggers.m.

  4. Confirm the PROPixx runs at 120 Hz (production presets refuse any other refresh rate) and that the projector desktop background is black.

  5. Check that both RESPONSEPixx button boxes respond (left box = left hand, right box = right hand).

  6. Set the sound level: in the two-tones launcher, use Play test tones with the participant in position and confirm the level is comfortable. Keep the Windows volume fixed for the whole session.

  7. Do a short Debug run of any task that changed since the last session.

Running the session

Start each task from its folder with main. In every launcher:

  • enter the same Subject and Session labels for all four tasks; the Run number is filled in automatically (run 1 for a new session);

  • keep the Production preset;

  • check the run length shown in the Planned run box, then press the Start button and confirm that the KIT recording is running;

  • press SPACE to start once the participant is ready and fixating.

1. Auditory: two tones

Auditory two tones (experiments/psychtoolbox/auditory_two_tones).

  • Passive listening to two short tones (500 Hz and 1000 Hz, 0.1 s, one every 1.2-1.6 s) while the screen stays static.

  • Defaults: 120 tones (60 of each): 2 min 50 s.

  • Instruct the participant to keep the eyes on the fixation point and simply listen; no response.

2. Artifact block

No stimulus script; the operator gives the instructions over the intercom and notes the clock time of each part. Record it as its own file (task artifacts). These recordings give clean examples of common artifacts for later artifact rejection and ICA.

  1. Tongue movements and swallowing: move the tongue around the mouth for a few seconds, then swallow a few times.

  2. Eye movements: look left, right, up and down a few times, keeping the head still.

  3. Rapid eye blinks: blink quickly several times.

Then measure mrk-2.

3. Motor: button presses

Cued left/right button press (experiments/psychtoolbox/button_press_lr).

  • A small arrow at fixation shows which hand to press with: left box = left hand, right box = right hand (any button on the box).

  • Defaults: 80 cues (40 per hand), cue 0.5 s, response window 1.5 s, one cue every 3-4 s: 4 min 42 s.

  • Instruct the participant to keep both hands resting on the boxes, look at the centre and press quickly with the indicated hand.

4. Visual categories: houses vs faces

Houses vs faces localizer (experiments/psychtoolbox/houses_vs_faces).

  • Passive viewing of 20 faces and 20 houses, each shown for 1 s (with a 0.25 s fade in and out), then 1 s of black with a fixation cross.

  • Defaults: 2 passes, 80 images (40 faces, 40 houses): 2 min 42 s.

  • Instruct the participant to look at the centre of the screen; no response.

5. Early visual: checkerboard

Flipped squares (experiments/psychtoolbox/flipped_squares).

  • In the Participant box, leave the epilepsy box ticked and the full-screen box unticked (see Photosensitivity).

  • After 2 s of grey, a small low-contrast checkerboard fades in over 1 s, reverses polarity twice per second, then fades out.

  • Defaults: 240 reversals: 2 min 07 s.

  • Instruct the participant to keep the eyes on the red dot; no response.

Sleep recording

Block 6: 60 min without stimulation, after mrk-3 and mrk-4, recorded as four 15 min recordings (sleep_1 to sleep_4) with one marker-coil measurement every 15 min.

  1. Close all experiment windows; leave the projector screen black (or turn the projector off) so no light reaches the participant.

  2. Dim the lights in the shielded room and make the participant comfortable for sleep. Tell them that they may fall asleep and that no task follows.

  3. Start sleep_1 and note the start time on the session sheet.

  4. Every 15 min: stop the recording, measure the marker coils, start the next sleep recording, and note the times.

  5. Monitor the participant on the camera and intercom throughout, and note any event (movement, waking, talking, staff entering) with its clock time and the time into the recording.

No triggers are sent during sleep, so the session sheet is the only record of these events.

Data

Stimulus PC

Each task writes BIDS files on the stimulus PC, below its own folder (data/ for production runs). For subject P01, session 01:

experiments/psychtoolbox/<task folder>/data/
  sub-P01/ses-01/meg/sub-P01_ses-01_task-<task>_run-01_events.tsv/.json
  sub-P01/ses-01/beh/sub-P01_ses-01_task-<task>_run-01_beh.tsv
  sourcedata/sub-P01/ses-01/beh/...   (snapshot, console log)

with <task> = twotones, buttonlr, housesvsfaces or flippedsquares. The KIT recordings (.con) and marker files (.mrk) come from the acquisition PC.

Project dataset

The project dataset is a BIDS dataset named epilepsy-project on NYU Box, below the folder given by the MEG_DATA environment variable (see the lab’s data storage conventions). One session per participant contains these recordings:

BIDS task

Runs

Events

twotones

run-01

events.tsv/.json from the stimulus PC

artifacts

run-01

none (times on the session sheet)

buttonlr

run-01

events.tsv/.json from the stimulus PC

housesvsfaces

run-01

events.tsv/.json from the stimulus PC

flippedsquares

run-01

events.tsv/.json from the stimulus PC

sleep

run-01 to run-04

none (times and events on the session sheet)

  • Use the same sub, ses, task and run labels for each converted KIT recording and its events.tsv/.json, which go next to it in the BIDS meg/ folder. The stimulus PC leaves onset and sample empty; they are filled from the recorded triggers.

  • Record which marker measurement belongs to each recording in the session’s sub-<id>_ses-<id>_scans.tsv (columns marker_before and marker_after); the pipeline’s KIT-to-FIF conversion uses this pairing. For sleep run-04 use the final mrk (see Session log).

  • The four tasks reuse KIT codes 224-229 for different events (224 is tone_low in twotones, cue_left in buttonlr and reversal_to_a in flippedsquares). Always read a code together with its run’s events.json.

Processing

The data are processed with the lab’s KIT general MNE pipeline (see the pipeline overview) using the project configuration epilepsy_project_config.yml:

  • the stimulus tasks go through bad-channel detection, ICA, filtering, events, epochs, source estimates and dipole fits (auditory N100m, pattern reversal P100m, faces/houses, motor fields of each hand);

  • no EOG channel is recorded, so the artifacts recording is the reference for what eye, blink, tongue and swallow components look like in this participant during ICA review;

  • the sleep runs get spectra and band power over time (delta to beta).

Processing of the project data in BESA Research is described on the BESA epilepsy pipeline page.

Per-session record

Record on the session sheet:

  • subject and session labels, date, staff present;

  • clinician approval reference and who was on call;

  • clock time of every mrk measurement;

  • for each block: KIT file name, start and end clock times, run number, anything unusual (ESC pressed, participant moved, sound level changed);

  • sound level setting used for the two-tones block;

  • for each sleep recording: start/end times, observed events with their time into the recording, and which mrk to use for co-registration.

Session log

First session (fill in subject, session and date):

Time

Block

Notes

4:58

mrk-1

5:02

Auditory: two tones

5:04

Artifacts: tongue / swallow, then eye movements, then rapid eye blinks

Time noted at the rapid eye blinks.

5:06

mrk-2

5:11

Motor: button presses

5:18

Houses vs faces

5:21

Checkerboard

5:22

mrk-3

5:25

mrk-4

after 5:25

Sleep, 60 min (sleep_1 to sleep_4)

One mrk every 15 min. sleep_4: sudden movement at about 230 s; co-register sleep_4 with the final mrk (measured after it).

Warning

Still to complete for this project: principal investigator, ethics approval number, inclusion/exclusion criteria, the clinical contact and the procedure if a seizure occurs during the session.

Data analysis

The ‘American Clinical Magnetoencephalography Society (ACMS) Clinical Practice Guideline 1: Recording and Analysis of Spontaneous Cerebral Activity’ provide a set of recommendations and obligations followed in the US when acquiring data from participant’s with Epilepsy and has been a basis used in this project.

Filtering

We follow the recommendation of the ACMS with applying a high-pass filter at 1 Hz and low-pass filter at 40 Hz.

Epileptic Activity

While this project is intended for research only, epileptic activity may appear on the MEG scan, as a change of rhytms or IIEDS (Interictal Epileptiform Discharges). Today, IIEDS are the main form of abnormal activity used in localizing the epileptic focus of the seizures.

If a seizure happens during an MEG scan (an ictal MEG scan), the onset of the seizure can be used to localize the focus of the Epilepsy. Since a seizure can spread fast to other areas of the brain, only the beginning

of the seizure activity is relevant for localizing the focus.

The localization technique of on the ictal onset should be well considered as it is not the same as for inter-ictal spikes.

Interictal spikes are also used to localize the focus of the epilepsy, these can be measured by MEG. Interictal spikes are averaged together to achieve a higher SNR then localized.

  • Spikes last 20-70 ms typically

  • Sharp waves last 70-200 milliseconds

Both are equivalently significant according to the ACMEG.

PCA and ICA can be used to estimate the number of signals participating in the signal that is not noise.

is IIEDS both spikes and sharp waves?

Choice of the specific points for source analysis

  • For source analysing of the IIEDS, points are picked from the rising part of the spike, or the peak of the spike.

MEG-Biomarkers for Epilepsy

What we are missing on

  • A note-taking synchronised with the MEG data recording: the acquisition member can take notes in real-time as the data is being acquired in synch with the recording

Questions

  • What is a good GOF when dipole fitting spikes?