MR Release Data🔗
MRI and MRS release data includes:
- Raw BIDS: Raw imaging/spectroscopy data in the standardized BIDS format
- Derivatives: Processed, analysis-ready imaging/spectroscopy pipeline outputs
- Tabular Imaging: Questionnaires, technician forms from scan session, and tabulated pipeline derivatives
Raw MR BIDS🔗
For MRI, DICOM images are converted using an HBCD-customized version of dcm2niix. For MRS, vendor-specific formats (Siemens `.dat`, Philips data/list, GE P-file) are converted using spec2nii v0.7.0. Additional post-processing is performed for MRI to ensure consistency across vendors. This includes converting QALAS to five 3D NIfTI files (labeled by inversion time) and hardcoding missing or inconsistent metadata fields for various modalities as follows:
| Field | DWI (Philips) | EPI (Philips) | BOLD (Philips) | T1w (Philips & GE) | QALAS (Philips, Siemens, GE) |
|---|---|---|---|---|---|
PhaseEncodingDirection |
|||||
TotalReadoutTime |
|||||
SliceTiming | |||||
{Small|Large}Delta | |||||
RepetitionTime | |||||
T2Prep | |||||
InversionTime |
Acquisition specs, extracted post-BIDS conversion from JSON metadata, must fall within the ranges specified below in order to be included in the release or any downstream pipeline processing. Note that inclusion criteria are typically defined as acceptable ranges rather than fixed values due to variations between scanner types.
| Scan Type | Repetition Time (TR) | Echo Time (TE) | Inversion Time (TI) | Slice Thickness | Number of Volumes |
|---|---|---|---|---|---|
| T1w | 2.3 - 2.41 | 0.002 - 0.0035 | 1.06 - 1.1 | 0.8 | NA |
| T2w | 2.5 - 4.5 | 0.09 - 0.15 | 0.29 - 0.33 | 0.563 - 0.565 | NA |
| MRS Localizer | 2.5 - 4.5 | 0.09 - 0.15 | 0.29 - 0.33 | 0.563 - 0.565 | NA |
| Diffusion | 4.8 | 0.0880 - 0.0980 | NA | 1.7 | ≥ 90 (AP + PA) |
| EPI Fieldmap | 8.4 - 9.2 | 0.064 - 0.0661 | 2 | 0.563 - 0.565 | NA |
| Functional | 1.725 | 0.0369 - 0.0371 | NA | 2 | ≥ 87 (~2.5 min) |
Each participant/session folder within rawdata/ contains raw imaging data organized into modality-specific subfolders (anat/, func/, etc.), with filenames encoding acquisition details (task, direction, run, etc.).
How To Read File Trees → hbcd/ └── rawdata/ └── sub-[ID]/ └── ses-[V0X]/ ├── anat/ │ ├── *_{T1w|T2w}.nii.gz # Structural MRI T1w & T2w │ ├── *_inv-{0-4}_QALAS.nii.gz # Quantitative MRI (QALAS) │ └── *_acq-mrsLoc{Ax|Cor}_run-{X}_T2w.nii.gz # MRS localizers │ ├── func/ │ └── *_task-rest_dir-PA_run-{X}_bold.nii.gz # Resting-state fMRI runs │ ├── fmap/ │ ├── *_dir-{AP|PA}_run-{X}_epi.nii.gz # fMRI fieldmaps acquired AP/PA for each run │ ├── *_acq-{anat|famp}_run-{X}_TB1TFL.nii.gz # qMRI Siemens B1+ fieldmaps │ └── *_acq-{tr1|tr2}_run-{X}_TB1AFI.nii.gz # qMRI GE/Philips B1+ fieldmaps │ ├── dwi/ │ ├── *_dir-{AP|PA}_run-{X}_dwi.nii.gz # Diffusion MRI │ ├── *_dir-{AP|PA}_run-{X}_dwi.bval │ ├── *_dir-{AP|PA}_run-{X}_dwi.bvec │ └── *_dir-{AP|PA}_run-{X}_sbref.nii.gz │ └── mrs/ # Acquired via shortTE and HERCULES ├── *_acq-{shortTE|hercules}_run-{X}_svs.nii.gz # Spectroscopy (SVS) └── *_acq-{shortTE|hercules}_run-{X}_ref.nii.gz # Reference scans
Derivatives🔗
Processed pipeline outputs are organized under specific folders named by processing pipeline as follows. See modality-specific subpages for details.
hbcd/
└── derivatives/
│ # Structural & functional pipelines
├── mriqc/
├── bme_x/
├── bibsnet/
├── nibabies/
├── freesurfer/
├── mcribs/
├── xcp_d/
│
│ # Quantitative MRI
├── symri/
├── qmri_postproc/
│
│ # Diffusion MRI
├── qsiprep/
├── qsirecon/
├── qsirecon-DSIStudio/
├── qsirecon-DIPYDKI/
├── qsirecon-TORTOISE_model-MAPMRI/
├── qsirecon-TORTOISE_model-tensor/
│
│ # MRS
└── osprey/
Structural and functional MRI derivatives with an average BrainSwipes QC score < 0.5 were flagged for expert manual review. Session data with confirmed severe structural issues were excluded from the release, including structural/functional pipelines outlined here, Osprey (MRS), and qMRI-PostProc. Sessions flagged for functional issues ONLY (at least one severely failing BOLD run) were only removed from Infant fMRIPrep and XCP-D derivatives.
Note that V02 sessions processed using Infant FreeSurfer (hash-2afa9081) for surface reconstruction should not be used for analysis - see Data Warning.
The table below reports the percentage of session folders removed due to QC failures for each visit.
| * Structural QC passed; one or more BOLD runs failed QC | ||||
| Visit | Surface Reconstruction Workflow | Structural Exclusions (%) | Functional Exclusions (%)* | Total |
|---|---|---|---|---|
| V02 | M-CRIB-S | 3% | 16% | 19% |
| V02 | Infant FreeSurfer | 19% | 30% | 49% |
| V03 | Infant FreeSurfer | 3% | 3% | 6% |
| V04 | Infant FreeSurfer | 0% | 3% | 3% |