(14/18) RBC data is also accompanied by a website to help facilitate data access and maintenance (reprobrainchart.github.io). If you need any help with RBC, find us on INCF NeuroStars using the RBC tag (neurostars.org/tag/rbc).

(12/18) Additionally, we found that higher overall psychopathology was associated with greater connectivity between the default mode and frontoparietal networks, suggesting a loss of normative developmental network segregation.

(11/18) The existing literature on associations between major dimensions of psychopathology and imaging features have been more varied. Our analysis of RBC data demonstrated that reduced SA and GMV (but not CT) were linked to higher overall psychopathology.

(10/18) Consistent with a rich literature from lifespan network neuroscience, we found evidence for a decline in between-network functional connectivity paired with increases in within-network connectivity, suggesting that resting-state networks become more segregated during development.

(9/18) To illustrate the utility of RBC, we also examined how neuroimaging features were related to participant age and overall psychopathology. Aligning with prior work, we found evidence for a decline in structural features during development following QC and harmonization.

(8/18) Emphasizing the “R” In RBC – every step of image curation and processing used DataLad and the “FAIRly-big” workflow to ensure full transparency and reproducibility – users can track all steps + rerun the pipelines as needed. MANY thanks to the amazing DataLad team!!!

(7/18) Data quality remains a very important confound in neuroimaging research, especially in neurodevelopment. All data in RBC went through a consistent QC process – huge credit to Nathalia B Esper + Alex R Franco! Good news is that the majority of RBC data have adequate quality!

(6/18) We used field-standard tools to process all the imaging data in a uniform way – FreeSurfer for structural images, and C-PAC for fMRI. We leveraged C-PAC’s configurability to have multiple processing streams – with and without GSR! – with LOTS of rich outputs.

(5/18) Another major challenge was that image acquisition parameters and image processing procedures vary considerably across studies. To characterize this heterogeneity & allow for fully reproducible curation, we developed CuBIDS & used it throughout RBC.
cubids.readthedocs.io

(4/18) Harmonizing psychiatric phenotypic data across studies was one of the hardest parts of the RBC project – heroes Mauricio S Hoffmann + Giovanni A Salum + Tyler M Moore came to the rescue! RBC provides a fully-harmonized bifactor model across all datasets.

(2/18) RBC aggregates phenotypic and neuroimaging data from five diverse and prominent developmental cohorts (N = 6,346) – the Brazil High Risk Cohort, the Chinese Color Nest Project, the Healthy Brain Network, the Philadelphia Neurodevelopmental Cohort, and the NKI-Rockland Sample.