SCAN Skill (Dataset-Orchestration Layer)

SkillDatabases & data

Use this skill for the Standardized Centralized Alzheimer's and Related Dementias Neuroimaging (SCAN) cohort, including NACC access planning, approved-export inventory, DICOM/NIfTI staging, BIDS organization, phenotype linkage, and MRI/PET processing. Trigger on 'SCAN dataset', 'NACC SCAN', 'ADRC imaging', 'SCAN MRI', 'SCAN PET', or requests to process a local SCAN export.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the SCAN Skill (Dataset-Orchestration Layer) skill

What this skill tells your AI

The instructions your AI receives, as published by cuhk-aim-group/neuroclaw in skills/scan-skill/SKILL.md and read by ahel’s review.

Overview

Use this skill to turn an approved SCAN export into an auditable multimodal workflow. Keep access control, identifier handling, staging, phenotype linkage, and modality processing explicit.

SCAN means Standardized Centralized Alzheimer's and Related Dementias Neuroimaging. It combines standardized MRI and PET from Alzheimer's Disease Research Centers with longitudinal NACC clinical and cognitive data.

Research use only.

Access gate

  • Request SCAN data through the NACC Quick Access File system.
  • Access is free but is not anonymous: NACC records a project proposal and approves the request.
  • Defaced images, QC fields, imaging summaries, and analysis variables are available through the request system. Availability varies because images must complete defacing and QC before release.
  • Do not automate account creation, approval, or authenticated scraping. Begin execution only after the user has placed an approved export on local storage.
  • Never ask the user to paste credentials into commands, logs, chat, or repository files.

Intake contract

Before processing, identify:

  • approved request or release identifier
  • export root and whether images are DICOM, NIfTI, or already BIDS-like
  • requested modalities and subject/session subset
  • NACC clinical/UDS tables supplied with the export
  • available stable identifiers and visit/date fields
  • local storage, compute, and DUA restrictions

Do not assume every participant has every optional sequence. Build modality availability from the files and QC tables, not from cohort-level descriptions.

Quick reference

TaskDelegate toOutput
Inventory approved exportclaw-shellimmutable file and checksum manifest
Convert approved DICOMdcm2niiNIfTI, JSON, bval, and bvec files
Stage or validate BIDSbids-organizerBIDS dataset plus validation report
T1w, T2w, or FLAIR processingsmri-skillstructural derivatives and QC
Resting/task fMRI processingfmri-skillfunctional derivatives and QC
dMRI processingdwi-skilldiffusion derivatives and QC
ASL/perfusion processingasl-skillperfusion derivatives and QC
Amyloid, tau, or FDG PET processingpet-skillPET derivatives, SUVRs, and QC
Join NACC phenotype tablesclaw-shell with structured tabular toolskeyed analysis table and join audit

Core workflow

  1. Inspect the approved local export without modifying source files.
  2. Create a manifest containing relative path, size, checksum, inferred modality, and source package identifier.
  3. Identify the subject and visit keys in both imaging and NACC tables. Preserve the source keys in a restricted crosswalk and generate separate BIDS-safe labels.
  4. Convert DICOM only when needed through dcm2nii; otherwise retain NIfTI and sidecars as received.
  5. Stage and validate the requested subset through bids-organizer. Record missing sidecars or acquisition metadata instead of inventing them.
  6. Join phenotype and imaging records using documented keys. Report unmatched, duplicated, and many-to-many records before analysis.
  7. Delegate only modalities actually present and approved for the project.
  8. Save QC, provenance, exclusions, and the exact cohort query beside the derivatives.

For a request that only asks whether SCAN is obtainable, stop after explaining the NACC access gate and do not claim that data were downloaded.

Identifier and privacy rules

  • Treat NACCID, PTID, visit dates, and crosswalks as controlled data even when images are defaced.
  • Keep source identifiers out of public filenames, reports, model artifacts, and example data.
  • Never attempt re-identification or facial reconstruction.
  • Do not redistribute source images, subject-level tables, or restricted derivatives.
  • Preserve the required SCAN/NACC acknowledgements and publication terms with the project provenance.

Expected output layout

scan_output/
|-- source_manifest/
|-- bids/
|-- phenotype/
|-- smri/
|-- fmri/
|-- dwi/
|-- perf/
|-- pet/
|-- qc/
`-- logs/

Completion criteria

  • The approved export and governing request are identified.
  • Checksums and modality inventory are saved.
  • Imaging-to-phenotype joins have explicit keys and an exception report.
  • BIDS validation and modality QC results are retained.
  • No restricted credentials, crosswalks, or source data are committed to Git.

References

Created At: 2026-08-11 HKT Last Updated At: 2026-08-11 HKT

Signals

GitHub stars
85
Forks
4
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
scan-skill
Source
github.com/cuhk-aim-group/neuroclaw