NIH Pediatric MRI (NIHPD)

The NIH MRI Study of Normal Brain Development, also called PedsMRI or NIHPD, is a multisite longitudinal study of typical brain maturation from the first weeks of life to early adulthood. It was run by the Brain Development Cooperative Group with funding from four NIH institutes (NICHD, NIDA, NIMH and NINDS). Healthy children were sampled to reflect the income and race or ethnicity of the US population, so the data can serve as normative controls for studies of childhood disorders and as test data for image analysis tools. The NIMH Data Archive collection 1151 holds Release 5.0 with 556 subjects that have shared data.

Composition

Data collection ran from 2001 to 2007. Objective 1 enrolled children aged 4 years 6 months and older and scanned them three times at roughly two-year intervals. Objective 2 enrolled younger children, from newborns to 4 years 5 months, and scanned them during natural sleep at intervals of three months to one year, up to ten visits. In the first cycle, usable structural MRI was obtained for 392 subjects aged 4.5 to 18 years and 72 subjects under 4.5 years. Each visit also collected demographics, neurological examinations, hormone levels, psychiatric interviews, cognitive tests and behavior rating scales.

Acquisition

Six pediatric study centers took part: Boston, Cincinnati, Philadelphia, Los Angeles, Houston and St. Louis. The core protocol had T1-weighted, T2-weighted and proton density-weighted scans. A subset received relaxometry, single-voxel and spectroscopic MRS, and DTI. DTI was acquired at five sites on 1.5T Siemens and GE scanners, with a short 3 mm protocol and an extended 2.5 mm protocol with several b-values up to 1100 s/mm². After quality review, 498 short and 193 extended DTI scans remain, from 274 and 152 subjects respectively.

Annotations

No manual labels are described. The study database pipeline included automated MRI segmentation, and the DTI processing center provides motion and distortion corrected diffusion data, tensor maps, quality scores and age-specific DTI templates.

Known limitations

• Access requires an approved NDA data use certification that must be renewed every year.
• The release counts per contrast, age and sex for the whole cohort are not stated in an openly accessible source; the main paper is not open access.
• Structural images and DTI data are in different spaces, and DTI raw files use the TORTOISE format.
• The DTI scanners are 1.5T systems, and all data date from 2001 to 2007.