Glossary Basics
Author: Virginie van Wassenhove virginie.van.wassenhove@gmail.com
Anonymization
To respect general ethical guidelines, participants should not be identifiable or recognizable. While it is rather easy to anonymize MEG or EEG data (simply removing metadata recording), it is more complicated for MRI because the face can be easily recognized. For this, MRI data have to be "defaced" (literally removing the face)
It is also recommended to remove the recording date and time from the files, and all other data that would allow a participant to identify themselves (date of birth, handedness).
Baseline (in MNE)
Baseline generally refers to the signals of reference e.g. the system at rest or unperturbed. Typically, when analysing event-related fields or potentials, the "baseline" is considered to be the pre-stimulus period (when no stimulation has been shown).
BIDS
Brain Imaging Data Structure
This is a world-wide community anonymizing, formatting and structuring of neuroimaging data for open-science. More information here.
Coregistration
After recording, you will need to spatially align the MEG data with an anatomical MRI recording (either of the same participant or a template). To do this, we load the digitized landmarks (+ additional digitized points) with the MNE coregistration tool and make sure the digitized points align with the anatomical landmarks on the MRI.
DICOMS
Format of aMRI files
Digitization
Typically used to record the fiducial coordinates in the MEG data. Additional points on the surface of the head are also collected to maximize the number of data points and inform realignment between M/EEG and aMRI.
Empty Room recording
Empty-Room are data recorded when none is in the room. This provides an estimate of baseline noise level in the room and the sensors on a specific day/time. We recommend recording empty room data right before you test your participant. These data can notably serve as a helpful noise covariance matrix during source reconstruction.
Metadata (in MNE)
In MNE-python you can attach a dataframe to your epochs object: it has as many lines as you have epochs/trials, and as many columns as you need to store additional information like reaction times, behavioral choices, or additional information about your stimuli.
More info here.
Fiducials (or markers)
In order to know where the participant's brain is with respect to the MEG device and to help realigning the MEG signals with the participant's MRI (coregistration), we digitize neuroanatomical landmarks which can be used post-processing.
The typical fiducial points are: the nasion (tip of the nose), the pre-auricular points (left and right little bump in the ear). Additionally, we can use a "multimodal" marker on the participant's cheek for instance.
SO: when you go and acquire MEG data, you will digitize (record) these points in the MEG data, and these will be used afterwards to align the MRI and the MEG.
Pipelines
We recommend writing your own code when processing your first datasets, to understand the process, but you can find examples in the CogBrainDyn MEG-pipeline, developed in collaboration between MNE python and our team, as a further development of the MNE Study Template. If your data are formatted in BIDS, you can also use the MNE-BIDS-Pipeline, which is a full-fledged processing pipeline for your MEG and EEG data.
Publishing Open-science
https://openneuro.org/datasets/ds003392/versions/1.0.4
Session
One recording (the full \~90 minutes being composed of several runs or blocks)
STIM channel
Name of a channel through which triggers are sent and recorded in the MEG data
Symbolic Link
(In linux) Pointer to the original file without copying the actual original file.
Triggers
Stimuli and participants' responses are being sent or collected by a computer that is separate from the MEG acquisition. Hence, the MEG data being recorded do not know what is happening when. To keep records or time-stamping of events directly in the MEG data, we send "triggers" which are basic signals sent from the PC stim to the MEG acquisition system.
In the .raw files (as stored by the machine during recording), these triggers are stored in a stimulus channel. MNE calls the triggers events and allows to associate names to them, for instance trigger 1 = 'visual event', 2 = 'auditory event', ..
BIDSification = renaming and organizing only the raw data in standardized way
Of course, after each pre-processing step (derivatives of raw data) the naming still requires to be done manually and the standard way of organizing is not yet developed!
Tutorials are here:
https://mne.tools/mne-bids/stable/index.html
Conversion example (with MNE sample subject):
https://mne.tools/mne-bids/stable/auto_examples/convert_mne_sample.html