How reliable are children's brain scans across different scanners?

How were the discovery and replication sets matched to ensure demographic comparability?

They were matched across 10 variables, including site location, age, sex, ethnicity, grade, parental education, handedness, family income, exposure to anesthesia, and family structure. Sibling and twin pairs were kept in the same set to avoid biases.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What strict head motion inclusion criteria were applied to the RSFC data?

Participants had to have an average framewise displacement of less than 0.20 mm across all neuroimaging runs, with at least 600 frames (8 minutes) of low-motion data. This high-standard motion censoring addresses systematic developmental bias.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What algorithm was used to derive the functional network organization of the group-average cortical surface?

The graph-theory-based Infomap algorithm for community detection was used. This approach grouped the averaged vertex-wise data into cortical functional networks.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

Which cortical parcellation scheme was utilized for the parcel-level RSFC analyses?

The study utilized a functional parcellation consisting of 333 regions of interest (ROIs) covering the entire cortical surface, defined by Gordon et al. This parcellation is known for high homogeneity of the BOLD signal within each parcel.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What was the correlation of the group-average resting-state functional connectivity (RSFC) matrices between the discovery and replication datasets?

The group-average matrices were nearly identical, exhibiting an extremely high Pearson correlation of r = 0.99. This demonstrates excellent group-level reproducibility.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

How was the spatial overlap of functional network assignments between the discovery and replication sets quantified, and what was the result?

It was quantified using normalized mutual information (NMI), yielding a value of 0.98. This indicates almost perfect spatial reproducibility of the network community structure.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What percentage of cortical vertices demonstrated different network assignments between the group-average ABCD child network solution and an adult network solution?

Overall, 29% of all vertices differed between the child and adult solutions. However, when excluding network borders, only 13% of the remaining vertices differed, indicating adult network topology is largely present by late childhood.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What percentage of the variance in resting-state functional connectivity (RSFC) was shared across all individuals in the sample?

Approximately 49% of the variance in RSFC was shared across individuals. This finding is highly similar to the amount of shared connectivity variance typically observed in adults.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

How did scanner manufacturer affect the similarity of resting-state functional connectivity (RSFC) patterns between participants?

RSFC was significantly more similar among participants scanned on Siemens systems than those scanned on Philips or GE systems. Significant differences in shared variance were found between manufacturers.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

In which brain regions was the effect of scanner manufacturer most prominent?

The effect of scanner manufacturer was most notable in posterior brain regions (such as the visual cortex), as well as in the anterior temporal lobe and orbitofrontal cortex. These areas are prone to distinct signal dropout and distortion profiles.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What kind of distance-dependent bias was observed in RSFC when comparing Siemens scanners to GE and Philips scanners?

A significant 'short-to-long' distance effect was observed, where short-distance correlations were stronger in Siemens scanners, while long-distance correlations were stronger in GE/Philips scanners. This bias could confound developmental interpretations if not controlled.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

Was there a difference in subject head motion across different scanner manufacturers?

Yes, participants scanned on Philips systems showed significantly higher average framewise displacement (FD = 0.13) compared to those scanned on Siemens (FD = 0.10) or GE (FD = 0.10) systems, indicating possible differences in the scanner noise floor.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

Did head motion differences account for the distance-dependent scanner manufacturer effects?

No. Even though Philips had higher head motion, the strongest distance-dependent RSFC discrepancy was between Siemens and GE scanners, which had no differences in head motion. This indicates that head motion does not explain the scanner effect.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

Did variations in altitude across data collection sites account for the observed scanner manufacturer effects?

No, the distance-dependent scanner effects remained significant and virtually unchanged after covarying for site altitude or completely removing high-altitude sites.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

How did the choice of head coil (32-channel vs. 64-channel) on Siemens scanners impact RSFC?

There was a small but significant difference, where RSFC was slightly more similar among participants scanned with a 32-channel coil. However, this effect size was substantially smaller than the overall manufacturer effect.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What was the effect of sex on global resting-state functional connectivity (RSFC) similarity in this cohort of 9-10 year olds?

The effect of sex was negligible to small, indicating that overall RSFC is highly similar between boys and girls at this pre-adolescent age.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

Which functional networks exhibited the greatest degree of inter-individual variability in functional connectivity?

Consistent with adult findings, the highest levels of between-participant variability were observed in networks related to top-down control and attention (frontoparietal and dorsal attention networks), along with the visual network.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

Was the high between-participant variability in the visual network driven by scanner manufacturer differences?

No, the high variability within the visual network was consistent across scanner manufacturers and not driven by scanner differences. It is hypothesized to reflect general physiological differences in arousal.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What connectivity patterns characterized the first principal component of between-participant functional connectivity variability?

It showed high positive loadings on connections within the visual and default mode networks, and broadly distributed connections between the default mode and dorsal attention networks. This component was significantly associated with cognitive ability.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What was the split-half reliability of the brain-behavior correlations across all connections between the discovery and replication datasets?

The split-half reliability was r = 0.60, demonstrating moderate and reproducible brain-behavior relationships when relating RSFC to general cognitive ability.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What statistical approach was adapted from genetics to correct for multiple comparisons across networks in the brain-behavior analyses?

The study used enrichment analysis, an analytical technique originally developed for genome-wide association studies (GWAS), which utilizes a hypergeometric test to assess the significance of network-level associations.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What instrument was used to assess individual differences in general cognitive ability?

The total "Cognition Composite" score from the NIH Toolbox cognition battery was used. This represents a highly reliable multi-domain measure of general cognitive ability.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

Which factors were included as covariates in the multiple linear regression models relating RSFC to general cognitive ability?

Scanner manufacturer and sex were included as categorical covariates. Scanner manufacturer was selected because it had a more appreciable effect on RSFC than data collection site.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What specific association between default mode network (DMN) and dorsal attention network (DAN) connectivity and cognitive performance was replicated?

Greater anticorrelation (negative functional connectivity) between the DMN (task-negative) and the DAN (task-positive) was significantly associated with higher general cognitive ability. This finding aligns with established models of cognitive control.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

How was connectivity between the frontoparietal network (FPN) and somatomotor hand (SMH) network related to general cognitive ability?

Stronger positive between-network functional connectivity between the FPN and the SMH was significantly associated with higher general cognitive ability in both datasets.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

Which within-network connectivity patterns were significantly associated with higher general cognitive ability?

Stronger within-network functional connectivity in both the default mode network (DMN) and the visual network was associated with higher general cognitive ability.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

How did the first principal component of between-participant RSFC variability relate to cognitive performance?

The first principal component of between-participant effects was significantly associated with greater cognitive ability in both the discovery and replication datasets, confirming that broad, multi-network patterns of variance track with intelligence.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

Were the observed brain-behavior correlations driven by an interaction with scanner manufacturer?

No, the network-level associations with general cognitive ability were not driven by scanner interactions; adding scanner interaction terms resulted in a negligible change in variance explained.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

Which image registration tool was used for the non-linear volume registration to standard space in the processing pipeline?

Advanced Normalization Tools (ANTs) non-linear registration was used to transform the volumes to a standard volume template space, improving on standard pipelines.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

How was head motion monitored in real-time during image acquisition at Siemens scanning sites?

Head motion was monitored online using Framewise Integrated Real-time MRI Monitor (FIRMM) software. This allowed operators to ensure at least 8 minutes of low-motion resting-state data were collected.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What major anatomical regions of the brain were excluded from the functional connectivity analyses in this study?

Subcortical regions and the cerebellum were excluded from these resting-state functional connectivity analyses, which represents a limitation of the study.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What precautions were taken during band-pass filtering in the "DCANBOLDproc" preprocessing stage?

The data was filtered between 0.008 Hz and 0.10 Hz without re-introducing nuisance signals or contaminating frames that were near high-motion censored frames.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

How were censored high-motion frames handled in the time series before denoising?

The processing pipeline performed interpolation across censored frames using least squares spectral estimation so that the data remained continuous for GLM denoising.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

What suggestion did the authors make regarding head motion thresholds for future multi-site studies?

They suggested that future studies might consider tailoring framewise displacement (FD) cut-offs on a per-scanner manufacturer basis, due to differences in scanner noise floors.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified

How did sorting the data by the 21 specific data collection sites impact RSFC similarity compared to scanner manufacturer effects?

Sorting by data collection site showed much smaller effects on resting-state functional connectivity (RSFC) similarity than sorting by scanner manufacturer, indicating successful site harmonization.

Source: Identifying reproducible individual differences in childhood functional brain networks: An ABCD study · DOI · quote verified