NIMH-IRP Data Science and Sharing Team collection

NIMH Data Archive collection 2936 is the release channel of the Data Science and Sharing Team of the NIMH Intramural Research Program. It bundles several MRI datasets collected in NIH intramural studies and shared as NDA Studies under one collection DOI. The bulk of it is the NIMH NHGRI data-sharing project, a structural MRI sample of healthy members of nuclear families from the National Human Genome Research Institute. It is useful for normative modelling and for image quality work; secondary analyses that built lifespan brain charts and a harmonization method draw on it.

Composition

The collection lists 447 shared subjects, all with records in the shared image data structure. 440 of them belong to the NIMH NHGRI release, whose description speaks of 441 people from 321 nuclear families, each with one T1-weighted scan. Two further releases from Peter Bandettini's group each contribute three subjects: a 3T task fMRI experiment with 100 runs per person, and a follow-up at 7T with 100 usable runs per person. A resting-state experiment is also registered. In the quality assessment paper that used the NHGRI scans, that cohort had 442 participants aged 5 to 78 years (mean 22.4), 252 male and 190 female, mostly children and adolescents.

Acquisition

The NHGRI T1-weighted images are MPRAGE scans with ASSET acceleration, 1.07 x 1.07 x 1.2 mm voxels and 124 slices, acquired on a 3T GE Signa scanner with an eight-channel head coil while participants watched a film. The task fMRI experiments used a GE 3T scanner and a Siemens 7T scanner with a 32-channel receive coil, with a visual stimulation and letter or number discrimination task and variants of it.

Annotations

Two raters scored every NHGRI T1 image for quality on a four-point scale, from good to unprocessable. Participants were free of psychiatric diagnoses according to clinician-administered structured interviews (SCID and the Conners adult ADHD interview for adults, the parent DICA for children).

Known limitations

This is a grouping of unrelated releases, not one designed study, and the collection page gives no per-release breakdown beyond the NDA Study cohorts. Family structure means many subjects are related. Demographics are published only for the paper's 442-person cohort, not the shared 440. The NDA page carries a notice that the repository is under review for possible changes.