ADNI T1w - CoBrALab/documentation GitHub Wiki

Reproducible BIDSifcation of ADNI

ADNI_to_BIDS

Downloading ADNI Datasets

1. Getting Access

1.1 Data Use Agreement

Before anything, read and agree to the ADNI Data Use Agreement:

1.2 Application

The application requires a brief research statement.

Timeline: Approval took less than a week.

2. Understanding ADNI Phases and Data

2.1 ADNI Phases

Phase Years Scanner Notes
ADNI 1 2004–2009 1.5T (main), 3T (~25%) T1 + dual-echo T2
ADNI GO 2010–2011 3T Bridge between ADNI 1 and 2
ADNI 2 2011–2016 3T Added FLAIR, T2*, accelerated T1s
ADNI 3 2016–2021 3T Added DTI, ASL, fMRI at all sites
ADNI 4 2022–present 3T Currently enrolling

2.2 Raw vs Preprocessed T1s

Collections of preprocessed images are provided by default. To access raw scans you must create your own "collection"

2.3 Data Advisory

Warning: All data with PTIDs matching the format 381_S_10### have been removed due to serious quality concerns. Do not use these in any analysis. See: https://adni.loni.usc.edu/news-publications/adni-data-advisory/

3. Navigating IDA

After login, the main interface is at https://ida.loni.usc.edu. Select ADNI as your project.

3.1 Key Areas

  • Search & Download β†’ Image Collections: For downloading imaging data (T1s, DWI, FLAIR, etc.)
  • Search & Download β†’ Study Data / Study Files: For tabular/clinical CSV data
  • Search & Download β†’ ARC Builder: Newer interface for tabular data β€” explore variables, build cohorts, download CSVs. Not used for imaging.

3.2 Pre-built Shared Collections

Under Image Collections β†’ Data Collections β†’ Other Shared Collections β†’ ADNI, there are pre-built collections like "ADNI1:Baseline 3T", "ADNI1:Complete 1Y 1.5T", etc.

4. Downloading Raw T1-Weighted Images

4.1 Using Advanced Search

Go to Search & Download β†’ Image Collections β†’ Advanced Search tab.

Set the following filters:

Filter Setting
Image Types Original only (leave Pre-processed and Post-processed unchecked)
Project/Phase ADNI 1, ADNI GO, ADNI 2, ADNI 3, ADNI 4 (select as needed)
Modality MRI
Acquisition Type 3D
Weighting T1

On the Image Description field: You can use *RAGE* as a wildcard filter to get all the MPRAGE (MP-RAGE) scans.

  • T1 + 3D + all descriptions: ~32,018 items
  • T1 + 3D + MPRAGE only: ~14,909 items

4.2 Duplicates Per Session

The results will contain multiple series per subject per visit. For example, a single visit may have:

  • MPRAGE (non-accelerated)
  • MPRAGE GRAPPA2 (accelerated)
  • Accelerated Sagittal MPRAGE

All are "Original" type. You will need to deduplicate after download, using QC files (see Section 6).

4.3 Creating a Collection

  1. From the Advanced Search Results tab, click Select All
  2. Click Add to Collection and give it a descriptive name (e.g., adni_all_t1_mprage_3d)
  3. Go to the Data Collections tab to manage and download your collection

4.4 Downloading

Click Advanced Download (top right of the Data Collections view).

4.5 Download Methods

Option A: Download Managers (Recommended)

The IDA Advanced Download page provides URLs that can be fed into a download manager. This is the best approach for large collections.

IDA-tested download managers:

Tool Platforms Site
jDownloader Linux, Mac, Windows https://jdownloader.org
Uget Linux, Windows https://ugetdm.com
FDM Mac, Windows https://freedownloadmanager.org

Workflow: Advanced Download β†’ copy the zip URLs β†’ paste into your download manager.

Option B: IDA Downloader (Java CLI Tool)

LONI provides an IdaDownloader jar file (found on the IDA support/download page). As of May 2026 the file is IdaDownloader_15May2026.jar.

Caveat: It requires Oracle Java. It will refuse to run on Ubuntu's OpenJDK.

Or if your HPC has modules:

module avail java
module load java/oracle-21  # or whatever Oracle JDK is available
java -jar IdaDownloader_15May2026.jar

Status: Not yet tested as of this writing, needs further investigation.

Option C: Browser Download

If the above don't work, fall back to the browser. Use Chrome, Edge, or Firefox.

Important constraints:

  • Same computer, same IP address for the entire download session
  • Sessions time out on inactivity don't let your machine sleep
  • Static IP required
  • browser page needs to stay open

What Doesn't Work

Tested and confirmed the following do not work:

  • curl with the download URLs directly, stalls at 0 bytes
  • curl -b cookies.txt with exported Firefox cookies (via cookies.txt addon)

4.6 Zip File Format

Downloads are in Zip64 format.

4.7 File Structure After Extraction

Images download as DICOM files in a deeply nested folder structure:

ADNI/
β”œβ”€β”€ 001_S_XXXX/
β”‚   β”œβ”€β”€ MPRAGE/
β”‚   β”‚   └──DATE/
β”‚   β”‚       └── SXXXXX/
β”‚   β”‚           └── *.dcm
β”‚   β”œβ”€β”€ Accelerated_Sagittal_MPRAGE/
β”‚   β”‚   └── ...
β”œβ”€β”€ 002_S_XXXX/
β”‚   └── ...

You will need dcm2niix to convert DICOM to NIfTI:

5. Downloading Tabular / Clinical Data

5.1 ARC Builder (newer)

Search & Download β†’ ARC Builder

Workflow: explore variables β†’ build cohorts with filters β†’ download.

6. Post-Download: Deduplication and QC

With raw downloads, you'll have multiple scans per subject per visit. To select the best one:

  1. Preferred scan: Check MAYOADIRL_MRI_IMAGEQC for the flagged "preferred" scan per visit
  2. Quality rating: Use MRIQUALITY.csv β€” ratings 1–3 are acceptable, 4 is failure/unusable
  3. Fallback: If no QC data exists, use the first (or most recent) scan for that visit

For ADNI 4, use the "chosen" series selected during Mayo ADIR Lab QC.

7. BIDS Conversion

7.1 Clinica adni-to-bids (Currently Broken)

The Clinica converter (https://aramislab.paris.inria.fr/clinica/docs/public/dev/Converters/ADNI2BIDS/) is the most established tool, but as of May 2026:

  • Not updated for the new LONI interface (broken since the Sept 2024 refactor)
  • Does not support ADNI 4
  • The Clinica team is waiting for the interface to stabilize before updating
  • Discussion: https://groups.google.com/g/clinica-user

7.2 Manual / Custom Approach

Without Clinica, the workflow is:

  1. Download raw DICOMs (Section 4) and clinical CSVs (Section 5)
  2. Use QC files to select one scan per subject per visit (Section 6)
  3. Convert selected DICOMs to NIfTI with dcm2niix
  4. Organize into BIDS directory structure:
bids/bash
β”œβ”€β”€ participants.tsv
β”œβ”€β”€ sub-ADNI001SXXXX/
β”‚   β”œβ”€β”€ ses-YYY/
β”‚   β”‚   └── anat/
β”‚   β”‚       β”œβ”€β”€ sub-ADNI001SXXXX_ses-YYY_T1w.nii.gz
β”‚   β”‚       └── sub-ADNI002SXXXX_ses-YYY_T1w.json
β”‚   β”œβ”€β”€ ses-ZZZ/
β”‚   β”‚   └── ...

8. Useful Links

Resource URL
ADNI main site https://adni.loni.usc.edu
IDA login https://ida.loni.usc.edu/login.jsp?project=ADNI
Data Use Agreement https://adni.loni.usc.edu/wp-content/themes/adni_2023/documents/ADNI_Data_Use_Agreement.pdf
MRI overview https://adni.loni.usc.edu/data-samples/adni-data/neuroimaging/mri/
MRI preprocessing docs https://adni.loni.usc.edu/data-samples/adni-data/neuroimaging/mri/mri-pre-processing/
Data advisory (bad PTIDs) https://adni.loni.usc.edu/news-publications/adni-data-advisory/
ADNI Data User Guide https://adni.loni.usc.edu/help-faqs/adni-data-user-guide/
Clinica converter docs https://aramislab.paris.inria.fr/clinica/docs/public/dev/Converters/ADNI2BIDS/
Clinica user group https://groups.google.com/g/clinica-user
FAQs https://adni.loni.usc.edu/help-faqs/faqs/
IDA User Manual (PDF) https://ida.loni.usc.edu/explore/jsp/support/IDA_User_Manual.pdf
IDA Support / FAQ https://ida.loni.usc.edu/explore/jsp/support/support.jsp