Derivatives & Preprocessing 

derivatives/
├── fmriprep/              # fMRIPrep v24.1.1 preprocessed data (all sessions)
├── fmriprep_nordic/       # fMRIPrep v24.1.1 run on NORDIC-denoised BOLD
├── nordic/                # Raw NORDIC denoising outputs (pre-fMRIPrep)
├── mriqc/                 # MRIQC v24.1.0 quality metrics and HTML reports
├── qc_review/             # HTML QC dashboards and BOLD QC benchmarks
├── atlases/               # Group-level reference atlases in template space
├── anatomical_rois/       # Subject-specific anatomically-defined ROIs
├── functional_rois/       # Subject-specific functionally-defined ROIs
├── hippunfold/            # HippUnfold hippocampal surface unfolding (whole-brain T2w)
├── hippunfold_oblcor/     # HippUnfold using oblique-coronal T2w acquisition
├── hsf/                   # Hippocampal subfield segmentation — HSF (Poiret et al.)
├── hsf_oblcor/            # HSF using oblique-coronal T2w acquisition
├── behavioral_analysis/   # Behavioral accuracy, d-prime, and learning analyses
├── bids_validation/       # Validation outputs, extracted event files, survey logs
└── fmriprep_pre022426/    # Archived earlier fMRIPrep run (legacy)

Target Pipeline 

sourcedata (DICOMs, raw behavioral)
    │
    ▼
BIDS raw (NIfTI + JSON + events TSV)
    │
    ├──▶ MRIQC (quality metrics, outlier detection)
    │
    ├──▶ fMRIPrep / fMRIPrep+NORDIC (preprocessing: registration, distortion correction, confounds)
    │
    ├──▶ atlases/ (reference parcellations in template space)
    │
    ├──▶ anatomical_rois/ (subject-specific anatomical segmentations)
    │
    ├──▶ functional_rois/ (subject-specific task-defined ROIs)
    │
    └──▶ ready/ (analysis-ready streams — see Analysis-Ready Preprocessing Pipeline)
  • MRIQC and fMRIPrep run in parallel, producing complementary QA output.
  • fmriprep_pre022426/ is an archived earlier run; the canonical output is in fmriprep/.
  • Analysis-ready outputs (ready/glmsingle/, ready/naturalistic/, ready/connectivity/) are documented in the Analysis-Ready Preprocessing Pipeline page.

Atlases

Group-level reference atlases stored at the template level (no per-subject data). Each atlas has a dataset_description.json and an atlas-<label>_description.json sidecar with provenance, license, and citation info.

Schaefer 2018

Local-global cortical parcellation (Schaefer et al., 2018, Cerebral Cortex). Available in 7-network and 17-network solutions at 8 granularities (100, 200, 300, 400, 500, 600, 800, 1000 parcels). All files are in MNI152NLin2009cAsym space at 2 mm resolution, matching fMRIPrep output.

derivatives/atlases/
├── dataset_description.json
├── atlas-Schaefer2018_description.json
└── tpl-MNI152NLin2009cAsym/
    └── anat/
        ├── tpl-MNI152NLin2009cAsym_atlas-Schaefer2018_seg-7n_scale-100_res-2_dseg.nii.gz
        ├── tpl-MNI152NLin2009cAsym_atlas-Schaefer2018_seg-7n_scale-100_res-2_dseg.tsv
        ├── ...  (7n × 8 scales + 17n × 8 scales = 32 NIfTI + 32 TSV)
        └── tpl-MNI152NLin2009cAsym_atlas-Schaefer2018_seg-17n_scale-1000_res-2_dseg.tsv

Entity key: seg-7n / seg-17n = network solution, scale-100 = number of parcels, res-2 = 2 mm isotropic resolution.

Anatomical ROIs

Subject-specific ROIs derived from structural imaging (e.g., hippocampal subfield segmentations, custom FreeSurfer-based masks). Stored per-subject (and optionally per-session if the definition is session-specific).

derivatives/anatomical_rois/
├── dataset_description.json
└── sub-03/
    └── anat/
        ├── sub-03_seg-hippsubfields_dseg.nii.gz     # Full subfield parcellation
        ├── sub-03_seg-hippsubfields_dseg.tsv         # Label lookup table
        ├── sub-03_label-CA1_mask.nii.gz              # Individual subfield mask
        └── sub-03_label-CA1_mask.json                # {"Type": "ROI", "Sources": [...]}

Functional ROIs

Subject-specific ROIs derived from task-based or resting-state analyses (e.g., localizer contrasts, seed-based connectivity maps). Each mask’s JSON sidecar documents the contrast, threshold, and statistical criteria.

derivatives/functional_rois/
├── dataset_description.json
└── sub-03/
    └── ses-01/
        └── func/
            ├── sub-03_ses-01_task-localizer_space-MNI152NLin2009cAsym_label-FFA_mask.nii.gz
            ├── sub-03_ses-01_task-localizer_space-MNI152NLin2009cAsym_label-FFA_mask.json
            └── ...

fMRIPrep (NORDIC)

fmriprep_nordic/ contains a parallel fMRIPrep run using NORDIC-denoised BOLD as input (see nordic/ below). Structure mirrors fmriprep/ exactly. The canonical source for each run (original vs. NORDIC) is recorded per-run in the QC decisions file; see Analysis-Ready Preprocessing Pipeline.

NORDIC

Raw outputs from the NORDIC thermal noise denoising step, applied to BOLD data prior to fMRIPrep. Each run produces a denoised NIfTI (.nii.gz) and a diagnostics file (.mat).

derivatives/nordic/
└── sub-03/
    └── ses-04/
        └── func/
            ├── sub-03_ses-04_task-TBencoding_run-01_bold.nii.gz
            ├── sub-03_ses-04_task-TBencoding_run-01_bold.mat
            └── ...

QC Review

HTML dashboards for visual QC of structural and functional scans, generated from MRIQC outputs. Includes per-subject and group-level views for T1w, T2w, and BOLD, plus bold_qc_benchmarks.md — a reference table of absolute IQM thresholds with citations.

derivatives/qc_review/
├── bold_qc_benchmarks.md
├── dashboards/
│   ├── qc_dashboard_all_bold.html
│   ├── qc_dashboard_sub-03_bold.html
│   └── ...  (per-subject T1w, T2w, BOLD dashboards)
└── ...

MRIQC

Image quality metrics generated by MRIQC v24.1.0 for structural and functional scans. Outputs include individual HTML visual reports and per-image quality metric (IQM) JSON files for automated outlier detection.

derivatives/mriqc/
├── dataset_description.json
├── logs/
├── sub-03/
│   ├── figures/
│   ├── ses-01/
│   │   └── anat/       # T1w, T2w quality metrics + figures
│   ├── ses-02/ ...
│   └── ses-30/
│       └── func/       # BOLD quality metrics + figures
├── sub-04/
├── sub-05/
├── sub-03_ses-01_acq-MPR_run-01_T1w.html   # Individual report pages
└── ...

Behavioral Analysis

Group-level and per-subject behavioral analysis results from the trial-based memory paradigm, generated by analyze_behavior.py in the mmmdata codebase.

derivatives/behavioral_analysis/
├── group/
│   ├── accuracy_by_enCon.tsv    # Accuracy broken down by encoding condition
│   └── dprime_by_subject.tsv    # Signal detection (d') per subject
├── figures/                      # Visualization outputs
├── sub-03/
├── sub-04/
└── sub-05/

BIDS Validation

Outputs from BIDS validation and event file extraction processes.

derivatives/bids_validation/
├── dataset_description.json
├── eventfiles/          # Extracted event files per subject
│   ├── sub-03/
│   ├── sub-04/
│   └── sub-05/
└── survey_logs/         # Pre-scan questionnaire processing logs

Analysis-Ready Outputs

The ready/ directory (GLMSingle, naturalistic, and connectivity streams) and the preprocessing_qc/ QC decisions files are documented in the Analysis-Ready Preprocessing Pipeline page.