FAQ
(Release 6.0)
Data access/use
Our data are publicly shared with eligible researchers with a valid research use of the data at a research institution. For 6.0 Data Release, please visit How to Access Data for more information on how to access and download the data.
5.1 and previous releases are handled by the NIMH Data Archive (NDA). The process for accessing the data starts by creating an account at NDA (if you do not already have one), and then requesting access to the ABCD Study data through the dashboard there (https://nda.nih.gov/user/dashboard/data_permissions.html).
Our data are publicly shared with eligible researchers with a valid research use of the data, and who are at an institution with an active Federal Wide Assurance, which many international institutions have (https://ohrp.cit.nih.gov/search/fwasearch.aspx?styp=bsc). Users from many other countries have successfully accessed and published with the ABCD Study data.
There is no cost to access the data.
All individuals can typically obtain their own data access, including graduate students, postdocs, and similar roles. However, it’s ultimately up to your institution to decide whether they will co-sign an independent DUC.
Group-Level DUC is generally recommended, where principal investigators (PI) submit a single DUC that includes trainees, lab members, and collaborators (adding or removing users is managed via user dashboard). The lead investigator is responsible for ensuring that individuals added as part of a group-level DUC are in compliance with its terms and conditions. Note that it is essential for the primary DUC holder to renew annually, as expiration will result in access being revoked for all listed collaborators.
Institutions vary on whether they consider use of the de-identified dataset to be human subjects research, with some requiring expedited-style or even exempt IRB reviews by their institution’s IRB. Other institutions do not consider it to be human subjects research given the de-identified nature of the data. That is a question to ask your IRB.
As a practice the ABCD Coordinating Center (CC) and Data Analysis, Informatics, and Resource Center (DAIRC) do not provide letters that could be seen to endorse specific applications or projects that propose secondary analyses of the ABCD Study data.
While we do not offer specific endorsements, we do offer our commitment to maintaining open lines of communication with the larger scientific community and to do whatever we can to ensure that they are able to acquire the information about the ABCD Data Resource that they may need to address their specific aims.
No, this is not permitted. Inputting ABCD data to generative AI tools, such as ChatGPT, is a violation of the terms of use outlined in the data use agreement.
Protocol
Protocol questions and response options are in our new interactive Data Dictionary (https:// data-dict.abcdstudy.org/). More information is on the ABCD Wiki: https://wiki.abcdstudy.org/. We are not able to share the actual surveys outside of the study, as many are proprietary. You can read more about the ABCD protocol development in a special issue of Developmental Cognitive Neuroscience here: https://www.sciencedirect.com/journal/developmental-cognitive-neuroscience/vol/32.
At this time, all analyzed specimens are part of the data release, but there is not yet a mechanism in place for the sharing of biospecimen remainders, and there is not currently an estimate of when that system might be in place.
There are no data on Autism diagnoses in the ABCD study, and we actually excluded children who already had confirmed Autism that was severe enough that they were not fully in main stream classes. Our exclusionary criteria for participation in the ABCD Study noted “Exclusionary diagnoses include a current diagnosis of schizophrenia, autism spectrum disorder (moderate, severe), mental retardation/intellectual disability, or alcohol/substance use disorder. (A past diagnosis that has remitted is not exclusionary).”
Imaging
Series that pass raw QC may have some minor issues but are considered acceptable for processing. Because of relatively tight brain coverage for dMRI and fMRI acquisitions, the superior or inferior edge of the brain is sometimes outside of the stack of slices. We term this field of view (FOV) cutoff. Except in extreme cases, dMRI and fMRI series are not excluded at raw QC for mild to moderate FOV cutoff. Such cases are also not recommended for exclusion by default. In the tabulated imaging data, brain regions outside the FOV have missing values, but other regions remain usable.
The automated post-processing QC metrics include measures of superior and inferior FOV cutoff that can be used to exclude participants with FOV cutoff from analyses. https://nda.nih.gov/data_structure.html?short_name=abcd_auto_postqc01
Raw gradient tables are available from the ABCD Collection page. Refer to specific notes on how to use these in the Fast Track Guidelines. In the minimally processed data, gradient tables are provided per scan, adjusted for head rotation.
The “minimally processed” (mproc) images have been corrected for B0 distortion using the field maps. As a result, field maps are not shared in the mproc data releases.
If you are obtaining fast track data (i.e., unprocessed), you would want to correct using the field maps provided in Fast Track.
To know which field map goes with a particular scan, here are some pointers:
- The study date and series time are included in the fast track file names.
- The protocol calls for two scans per task-fMRI session (i.e., two for MID, two for SST, and two for nBack).
- Each pair scans is typically proceeded by a field map scan (or pair of field map scans).
- On GE scanners, the field map is an integrated sequence of forward and reverse phase-encode polarity scans (i.e., one field map scan for GE). On Philips and Siemens scanners, the forward and reverse scans are separate series (i.e., two field map scans for Philips and Siemens).
Here are some extra things to know:
- Sometimes field map scans have image quality issues and may fail raw QC. Those will be excluded from Fast Track downloads if you access the “recommended” Fast Track query.
- If there is no usable field map scan directly preceding a scan or pair of scans, we use the next closest field map in the imaging session.
- Regardless of whether the field map is collected immediately before the pair of tasks scans, or earlier, or later, it is advisable to correct for motion between the field map and the main scan.
- If you use FSL’s TOPUP, it corrects for motion between the forward and reverse phase-encode polarity volumes, which occurs in many subjects as well.
Sorry, FreeSurfer derivatives are not included in data shared via NDA, only “minimally processed” image volumes.
The ABCD imaging protocol is described in Casey et al., 2018, The Adolescent Brain Cognitive Development (ABCD) study: Imaging acquisition across 21 sites. Dev Cogn Neurosci. 32:43-54. doi: 10.1016/j.dcn.2018.03.001. Epub 2018 Mar 14. Review. PMID: 29567376.
For imaging data from Philips scanners, the dMRI acquisition is split into two series because of a limitation of the Philips platform. Both scans have the same phase-encode polarity. They are meant to be concatenated together. In other rare cases, multiple dMRI scans may have been acquired, due to acquisition problems in early scans. For the minimally processed data, one scan is selected for each session based on QC ratings, except for Philips scanners, in which case two are selected for packaging and sharing. All scans are available as raw DICOM files via fasttrack data sharing.
Minimally processed neuroimaging data includes:
- High-resolution structural data (3D T1w and T2w scans)
- Advanced diffusion MRI (multiple b-values and directions)
- Resting-State fMRI
- Task fMRI (Monetary Incentive Delay, Stop-Signal, and Emotional N-Back), with event files for each fMRI run.
These series have been run through standard modality-specific pre-processing stages including conversion from raw to compressed files, distortion correction, movement correction, alignment to standard space, and initial quality control (refer to MRI Quality Control (QC) Release Notes). This is to enable researchers to use the ABCD neuroimaging data in their own processing pipelines more quickly and efficiently than starting with raw data. Note that minimal processing is identical for rs-fMRI and task-fMRI and does not include analysis-specific pre-processing steps (e.g. removal of initial TRs, normalization by mean, etc.). Researchers intending to use minimally processed data should take note of the appropriate acknowledgment language to include in any public disclosure of results (refer to https://data-archive.nimh.nih.gov/abcd/results). The available minimally processed files are detailed in the Other Imaging Instruments Release Notes.
Preprocessed imaging data are packaged in archive files (tgz) for each image series containing BIDS formatted directory trees and NIfTI format data files (software to share preprocessed data: https://scicrunch.org/resolver/SCR_016016; consistent with BIDS specifications version 1.1.1: http://bids.neuroimaging.io/bids_spec.pdf). Imaging metadata derived from the original DICOM files are packaged along with each preprocessed data series as JSON files. The minimally processed T2w data are resampled into voxel-wise alignment with the T1w, which is rigid-body resampled into alignment with an atlas.
dMRI-specific information included diffusion gradients adjusted for head rotation (bvecs.txt), diffusion gradient strengths (bvals.txt), and a rigid-body transformation matrix specifying the registration between the dMRI image and the corresponding processed sMRI T1w image (stored in the JSON file). The dMRI minimally processed data are also kept in their original resolution, but reoriented into a standard alignment, based on registration to T1w, but not voxel-wise aligned with the T1w. A registration matrix supplied with the minimally processed dMRI data.
fMRI-specific information includes estimated motion time courses and a rigid-body transformation matrix specifying the registration between the fMRI image and the T1w image (stored in the JSON file). The fMRI minimally processed data are kept in their original space and resolution, but a registration matrix is supplied with the minimally processed fMRI data. For task-fMRI series, event timing information is included as tab-separated value (tsv) files. The results of additional processing and ROI analysis are shared in tabulated form to the NDA database (https://scicrunch.org/resolver/SCR_016010), from which users can export spreadsheet files (tsv).
Information about this is included in the release notes and in our recent publication, Hagler et al., 2019, NeuroImage. Image processing and analysis methods for the Adolescent Brain Cognitive Development Study (doi:10.1016/j.neuroimage.2019.116091). They also describe what processing steps are included in the “minimally processed” data shared on NDA.
There is currently no script available to run the ABCD minimal processing. There is a Docker that runs the complete processing and analysis pipeline available at https://www.nitrc.org/projects/mmps_docker/. Other useful software packages include https://github.com/ABCD-STUDY/abcd-dicom2bids and https://github.com/ABCD-STUDY/abcd-hcp-pipeline.
Please refer to the Imaging release notes. There are modality-specific image inclusion flags, like imgincl_t1w_include or imgincl_rsfmri_include, that are either 0 or 1. The fsqc_qc variable is used in the inclusion criteria to set the inclusion flags (criteria are listed in the release notes).
As described in the release notes, not all visits had manual post-processing QC; instead a sampling approach was used. For Release 3.0, post-processing QC (including FreeSurfer QC) was done for ~5-10% of visits for each modality. All visits that had failed QC in Release 3.0 were re-reviewed for Release 4.0. Because FreeSurfer version was unchanged between Release 4.0 and 5.0, reviews from Release 4.0 carried forward to Release 5.0. Additionally for Release 4.0 and 5.0, as described in the release notes, we selected additional visits for manual review based on automated QC metrics and multivariate outlier detection.
Please refer to the ABCD Release Notes: Imaging Instruments. Image processing and analysis methods corresponding to ABCD Release 2.0.1 are described in Hagler et al., 2019, Image processing and analysis methods for the Adolescent Brain Cognitive Development Study. Neuroimage, 202:116091. Changes to image processing and analysis methods are documented in the relevant release notes.