³ÉÈËVRÊÓƵ

Purpose

OMEGA is amongst the largest repository of MEG data presently available. It is a continuously expanding resource that also contains anatomical MRI volumes, demographic and questionnaire information. OMEGA may also feature multimodal electrophysiological data (e.g., EEG, field and cell recordings).

The primary objective of OMEGA is the conservation of brain imaging data collected with different brain imaging techniques and behavioral tests scores from healthy and patient volunteers (any age range), in order to allow future research in the field of high-powered detection of effects of interest in human brain activity.

OMEGA is a resource to establish typical features of brain activity, in relation to demographic factors (age, gender, standardized behavioral tests scores, etc.) and to reveal new markers of abnormal brain activity in various pathologies.

We hope OMEGA contributes to open scientific research, for the advancement of knowledge, and the greater good.

---

Highlights

• First open data repository fully dedicated to MEG.
• MEG data from 644 participants (see details below): 444 healthy controls and 200 patient volunteers (Parkinson, ADHD, chronic pain). About 1800 resting-state MEG recordings totaling over 150 hours of data.
• Structured according to the (v1.10.0).
• Directly readable by data-analysis software like .
• Anatomical T1-weighted MRI volumes (defaced) of most participants, with co-registration now available for many (work ongoing).
• Demographic and standardized questionnaire data.

Participants

Group	Sex	18 - 19	20 - 29	30 - 39	40 - 49	50 - 59	60 - 69	70 - 79	80+	Tot
Control	F	12	109	23	9	14	78	26	2	273
Ìý	M	12	83	19	7	3	33	12	2	171
Parkinson	F	Ìý	Ìý	Ìý	2	16	12	13	3	46
Ìý	M	Ìý	Ìý	Ìý	8	23	50	31	3	115
ADHD	F	4	7	2	Ìý	Ìý	Ìý	Ìý	Ìý	13
Ìý	M	4	4	Ìý	Ìý	Ìý	Ìý	Ìý	Ìý	8
Chronic Pain	F	Ìý	Ìý	1	1	3	2	1	Ìý	9
Ìý	M	Ìý	Ìý	Ìý	3	5	1	Ìý	Ìý	9
Total	Ìý	32	203	45	30	64	176	83	10	644

Current version

The present version is Release 3.1.1, 2025-07-17: it added co-registration for 480 participants. Version 3.1 re-introduced most participants shared previously, as well as more new data from recent studies, totalling over 600 participants!

Next releases

• Completing manual co-registration of MEG & MRI for remaining subjects
• Partial medication list for PD patients
• Sleep data

Access

OMEGA promotes open-science good practices via, presently, a controlled access procedure.

You will need to provide your full name, affiliation, and a valid institutional email address (Gmail, Yahoo, iCloud, etc. addresses are not permitted). This information is for verification and curation purposes only, and will not be used outside of OMEGA.

You are also asked to review and accept the following terms of the OMEGA Data Use Agreement:

• You must agree to respect the data privacy standards in effect in Quebec and Canada. In particular, you must not attempt re-identification of participants nor redistribute the data. Please redirect any request to access your copy of OMEGA to the present page.
• You must acknowledge that the proposed study has been reviewed and approved by the duly constituted research ethics board(s) that has/have jurisdiction over it – typically the IRB/REB from your home institution. However, if the project originates from a jurisdiction where ethics approval is not required, you must seek and have obtained approval from , and provide the approved REB study number at the time of your registration to OMEGA.

• You will need to keep the retrieved OMEGA data in a secure environment. You shall not share this copy or use it to create a new bank or repository.
• You will recognize the contribution of OMEGA in all your publications and presentations of study work or results obtained using data from OMEGA.

Ìý

Please click here to register.

Data

Retrieving data

The data is stored on a secure Nextcloud instance at the BIC (migrated from our previous ownCloud server in Dec 2025). is an open-source web-based file hosting application similar to Dropbox or Google Drive. Once you have been granted access, you will receive access credentials by email. You can then download files via the following options. Additional configuration details are available on the .

1. is simple, but not the most efficient for downloading large volumes of data at once.
2. can be slightly more complex to set up, depending on the OS, but is a versatile option. It allows mounting the shared directory and copying files with common file managers (e.g., Windows Explorer, Mac OS X Finder, Linux KDE Dolphin) or from the command line, though this may give slow speedsÌýin some configurations. You may try a WebDAV client (e.g., or ) to help with the setup and provide a user-friendly interface to retrieve files. It is also possible to download directly from the command line, e.g. in an HPC environment, with tools like wget, curl, or rclone. Our currently preferred method under Ubuntu is rclone.
3. is designed to synchronize files between a local computer and aÌýNextcloud server. It has not been tested by us yet to see if it can beÌýused by guests to download shared files.

Data organization

OMEGA is structured according to 1.10.0.

Apart from some general files in the root folder (dataset_description.json, participants.tsv, CHANGES, LICENSE, etc.), and the demographic information in the phenotype subfolder, the imaging data is organized by participant, session and modality, as the following example:

sub-0000/
ÌýÌýÌýÌýses-01/
ÌýÌýÌýÌýÌýÌýÌýÌýsub-0000_ses-01_scans.tsv
ÌýÌýÌýÌýÌýÌýÌýÌýanat/
Ìý Ìý Ìý Ìý Ìý Ìý sub-0000_ses-01_T1w.nii.gz
ÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýsub-0000_ses-01_T1w.json
ÌýÌýÌýÌýÌýÌýÌýÌýmeg/
ÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýÌý sub-0000_ses-01_task-noise_run-01_meg.ds
ÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýsub-0000_ses-01_task-noise_run-01_meg.json
ÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýsub-0000_ses-01_task-noise_run-01_channels.tsv
ÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýÌý sub-0000_ses-01_task-rest_run-01_meg.ds
ÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýÌý sub-0000_ses-01_task-rest_run-01_meg.json
ÌýÌýÌýÌýÌýÌýÌýÌýÌýÌýÌý sub-0000_ses-01_task-rest_run-01_channels.tsv
Ìý Ìý Ìý Ìý Ìý Ìý sub-0000_ses-01_task-restaftertask_run-05_meg.ds
Ìý Ìý Ìý Ìý Ìý Ìý sub-0000_ses-01_task-restaftertask_run-05_meg.json
Ìý Ìý Ìý Ìý Ìý Ìý sub-0000_ses-01_task-restaftertask_run-05_channels.tsv
Ìý Ìý Ìý Ìý Ìý Ìý sub-0000_ses-01_coordsystem.json
Ìý Ìý Ìý Ìý Ìý Ìý sub-0000_ses-01_headshape.pos

In each session folder, a _scans.tsv table lists the MRI scans and MEG recordings available.

MEG data

OMEGA contains data collected as part of many different studies. A few of the larger projects collect additional data such as clinical info, neuropsychological tests, imaging, genetics, etc. that are shared in separate repositories. In these cases, participant IDs in OMEGA match the IDs in the other repositories, or a key for matching IDs can be provided upon request. This is currently the case for these two projects:

• Ìý(Pre-symptomatic Evaluation of Experimental or Novel Treatments for Alzheimer's Disease)
  Participant ID:Ìýsub-CONP0000 (119 participants)
  Access:ÌýÌý
• Ìý& Canadian Open Parkinson Network Database; and
  The Neuro Open Biobank (aka Clinical Biospecimen Imaging and GeneticÌýRepository, C-BIG)
  Participant ID: sub-PD0000 and sub-MNI0000 (157 participants)
  Access:ÌýÌý& ,Ìý

Currently, OMEGA only contains resting-state MEG recordings. Task recordings may be added in future releases. "Baseline" rest recordings are identified by the task label "rest". If a rest recording was collected after the participants performed a task, the label is "restaftertask".

Empty-room "noise" MEG recordings were collected for most MEG sessions and are included in each session folder, next to the participant's recordings.

The data were all performed with the 275-channel 2005 series system at the McConnell Brain Imaging Center.

Raw MEG data files were renamed, de-identified and structured according to BIDS with custom MATLAB code, some of which is available in our (2024-10-10).

When system issues, environmental noise, or participant-specific artefacts (e.g., dental work or makeup) were noted by the individual acquiring the data, this information is saved in the _meg.json file, in the SubjectArtefactDescription field. We will update this information as it is reported by users or identified by improved quality control procedures.

Most recordings are resting state in sitting position, with eyes open, and with EOG and ECG electrodes. However a few studies had slightly different setups. While this information can be found in the files of each recording, these special cases are also listed in the README.md file for convenience.

MRI data

One or more T1w MRI images are included for each participant, when possible, for source imaging purposes.

Raw DICOM images were sorted, de-identified and converted to the NIfTI-1 file format and BIDS structure with (0.10.0). Images were then de-faced (facial features replaced by a template to further preserve anonymity) with (0.3.3).

Co-registration

MRI-MEG alignment information is now provided for many participants (work in progress). See Co-registration tab.

Scan dates

As an additional anonymization measure, session dates were shifted by a fixed number of days within each participant volume (the shifting procedure sets the first sessions for each participant to 2000-01-01.) Hence, the relative date difference between sessions for each participant remains meaningful. These relative times can also be obtained with the (fractional) age provided in the sessions table (see demographics section below).

OMEGA general questionnaire

The questionnaire administered to participants evolved over time, therefore the available data varies between subjects. General information such as age, gender, sex, handedness, etc. is available for the bast majority of participants. All questionnaire versions are now combined in a single table.

Some responses from previous versions were recorded in French and still appear as such in the table. We aim to translate most answers in an upcoming release.

The questionnaire is primarily a collection of brief standardized tests. It is described in detail in the file phenotype/OmegaQuestionnaireDescription.pdf, and the questions themselves are provided in the accompanying json files (English and French).

Participant demographics

Basic participant information is provided in the file participants.tsv. As some volunteers participated in multiple studies and their age or category may change between sessions, an additional table is provided with per-session values: phenotype/sessions.tsv. The age and category in the participants table correspond to the first MEG session for each participant.

MRI-MEG Co-registration

We have co-registered the MEG and MRI data for most participants, to best align anatomical and functional brain data and maximize MEG source mapping accuracy. The present page details the co-registration process, how the related information is saved in OMEGA, and introduces a quality scoring system to assess the outcome of the co-registration procedure. Data processing and visualization were performed with Brainstorm (Tadel et al. 2011), which supports both automatic and manual registration. Custom scripts were also used.

Note: Brainstorm now automatically loads co-registration information when importing OMEGA as a BIDS dataset.

Motivation

Due to privacy constraints, MRI data shared in OMEGA are defaced, and photos of MEG head coil placements cannot be shared. Additionally, coil placement conventions and digitization quality vary across contributing studies. As a result, automatic alignment based on digitized head points, as described in , is often unreliable. We therefore provide precomputed co-registrations using original (non-defaced) MRI images, digitized head shapes and fiducials, session notes, and head coil placement photos, as available for each MEG session.

Procedure

Accurate source localization in MEG requires that the sensor and MRI data be represented in the same 3D space. This is typically achieved through rigid transformation (translation and rotation) based on corresponding anatomical landmarks in both coordinate systems. Several methods are used:

1. Using MEG Head Coils

This method uses the three MEG head coils, localized by the system at the start of each recording, as reference points. Their MRI positions must be manually annotated using photographic documentation or standardized conventions. While conceptually simple, this method is time-consuming and error-prone due to the difficulty of identifying coil locations in MRI slices.

2. Using Digitized Head Points and Fiducials

This approach relies on a 3D digitizer to record scalp surface points and landmarks. These points can be automatically aligned to the MRI-derived head surface. While this method provides high accuracy when the digitization is of good quality, it is constrained by rigid transformation of the entire point set and less flexible than coil-based alignment.

3. Manual Coil Alignment Without Digitization

For sessions lacking reliable digitization, we plan to use the MEG-localized head coil positions for manual alignment onto the MRI-derived head surface. This method benefits from visual reference to the 3D surface, preserves the relative geometry of the coils, and avoids navigating through MRI slices.

Digitization quality

Most digitization in OMEGA was performed using a Polhemus Fastrak system, which relies on a reference receiver affixed to the head. If this receiver shifts during acquisition, the resulting head shape becomes distorted. In such cases, scalp points are ignored, and only digitized coil positions are used for alignment, assisted by session photographs. If the relative geometry of the digitized coils is compromised, even this strategy becomes infeasible, necessitating full manual alignment without digitization.

File Format and Storage

OMEGA follows the (BIDS) specification for co-registration. Landmark coordinates in both MRI and MEG native spaces are saved under AnatomicalLandmarkCoordinates field in the MRI _T1w.json file and the MEG _coordsystem.json file.

These files are located in the subject's anat and meg session folders. The IntendedFor field in the _coordsystem.json file links each MEG session to the MRI used for co-registration.

MRI coordinates are stored in 0-indexed voxel space; MEG coordinates use the CTF system (in cm), defined by head coil positions.

Co-registration status is reported in phenotype/sessions.tsv, with coregistration (Yes/No) and coreg_confidence (0–3) columns.

Multiple Sessions

BIDS supports two options for linking multiple MEG sessions to a single MRI: (1) session-specific landmarks listed in the MRI _T1w.json, or (2) reuse of MRI landmarks across sessions. OMEGA adopts the latter. The MRI landmarks are typically digitized fiducials from the subject’s first MEG session, often anatomical points, but may correspond to MEG coil positions in older data. The labels (NAS, LPA, RPA) are retained consistently, regardless of origin.

Note: Landmarks in _coordsystem.json may not match those in _headshape.pos, which should be ignored when co-registration is provided. The FiducialsDescription field in _coordsystem.json provides further detail on landmark sources.

For participants with multiple anatomical scans, the MRI used for co-registration is noted under IntendedFor. Additional MRIs are retained but not curated for quality. In Brainstorm, only one MRI is used to define the anatomy of each subject. When importing a BIDS dataset such as OMEGA, scans with co-registration metadata are prioritized.

Confidence Rating Scale

Each co-registration is assigned a quality score from 0 to 3. Ratings are based on quality of the fit between digitized points and head surface, and the availability of validation photos. Definitions are also provided in README.md and phenotype/sessions.json:

• 3: High-quality. The 3 digitised head coils appear where expected on the MRI head surface, which is confirmed with session photos. Head points closely fit the MRI surface (≤5 mm), with no systematic error.

• 2: Acceptable. Minor discrepancy in coil alignment or head point fit, validated with photos. If no photos are available, digitized coils and head points appear very well aligned with the head surface.

• 1: Low-quality. Major alignment issues with digitized coils compared to photos, even ignoring other head points. If no photos are available, alignment of digitized coils and head points with head surface appear acceptable, with no systematic error.

• 0: Not co-registered. No usable MRI, missing or unusable digitization, or irreconcilable errors in coil position geometry.

Current Status and Future Work

As of release 3.1.1, co-registered MRI data are available for 479 subjects (first MEG session only), based on digitized head shapes and fiducials. Work is ongoing to extend co-registration to all sessions and subjects with MRI data.

Sessions rated 0 or 1 will be revisited using the MEG-localized coil positions rather than digitized points to improve alignment quality.

Contact

For any inquiry, please contact , OMEGA Principal Investigator, at sylvain.baillet [at] mcgill.ca ()