Generation R Study (neuroimaging)

Generation R is a prospective birth cohort run by Erasmus MC in Rotterdam, the Netherlands. It enrolled 9,778 pregnant women with due dates between April 2002 and January 2006 and follows their children from fetal life onward. Its neuroimaging arm scans the children's brains at repeated visits so that brain development can be linked to prenatal exposures, genetics, cognition and behaviour measured in the same families.

Composition

The first imaging wave started in September 2009 and scanned 1,070 children aged 6 to 9. The second wave started in 2013: 4,245 children came to the MRI centre, 4,087 were scanned and 3,992 had valid parental consent for research. Of these, 3,959 have a complete T1-weighted scan, 3,687 a T2-weighted scan, 3,777 diffusion imaging and 3,439 resting-state fMRI. The second-wave sample is 1,975 boys and 2,017 girls; 41 have a clinically confirmed autism spectrum disorder diagnosis. 640 children have T1-weighted scans from both waves. A third wave around age 13 was ongoing in 2017, with over 800 children scanned at that time.

Acquisition

Wave 2 used a dedicated GE Discovery MR750w 3 T scanner, mostly with an 8-channel head coil and the DV24 software. The protocol had a 3D inversion recovery fast spoiled gradient echo T1-weighted scan at 1 mm, a 3D T2-weighted fast spin echo, single-shell diffusion with 35 directions and 3 b=0 volumes, and a 6 minute resting-state fMRI run of 206 volumes. Children practised in a mock scanner first and watched a film during structural and diffusion scans. Wave 1 also used a GE MR750 3 T system. The cohort update lists a 2D proton-density turbo spin echo, which the wave 2 protocol table does not include; it is left out of the contrast list.

Annotations

T1-weighted scans were rated for quality at the scanner and a random 500 were rated on a structured scale. Trained raters screened scans for incidental findings, with neuroradiologist review of abnormal cases.

Known limitations

• Images are not public. Access runs through a collaboration request reviewed by the study management team and the medical ethics committee, and no data use terms are published.
• Families who joined the imaging waves were more often Dutch, higher-income and more educated than the full cohort.
• The two design papers disagree on the wave 1 size (1,070 versus 801); the neuroimaging paper's number is used.