Thread by @GaabLab

I'll summarize a recent discussion focused on some of the main challenges in developmental cognitive neuroscience, especially for young participants & longitudinal data. Please add to this & maybe we can turn it into a resource for the field 1/11 #mri #eeg #nirs #devneuro #brain

1) Vascular differences: Does the difference in the vascular structure between infants/toddlers/older children lead to problems with interpreting/analyzing #dedvneuro data? Do we need age-specific #hrf functions? What about circadian rhythm effects? 2/11

2) Parcellations/templates: Which #template should you use when analyzing longitudinal data, especially if the brain changes a lot over the study's time course (e.g. age 0-5). Is 'template fit' a confound in these studies? Do we need age-specific parcellations? 3/11

4) Segmentations: Several labs are developing segmentation tools. However, segmentations are more accurate in older children than in newborns and infants. How can we control for this in longitudinal studies? 4/11

4) ROIs (structural or functional): How do we select ROIs across longitudinal time points if the #brain is changing a lot (e.g. from infant to toddler to preschooler)? Does vascular and structural maturation introduce confounds here? 5/11

5) Tasks/task duration: How can we design longitudinal experimental tasks that enable compliance in young children & older children & give us reliable data & enough data points. For example, very young children often do better if tasks are separated in different runs. 6/11

6) Longitudinal studies/Sequences: Should one keep the same sequences over many years in a longitudinal study or update sequences? How can this potentially affect data collection/analysis? How should we handle scanner updates/retirements in longitudinal data sets? 7/11

7) Skull thickness/hair color/thickness: Are these confounds for EEG/NIRS? How should we deal with these confounds over longitudinal time points? How does impedance change over time? 8/11

8) Motion: How do we deal with differences in motion across longitudinal time points. Younger children move more than older children which can introduce a systematic bias. Do we introduce age-specific motion cut-offs? 9/11

9) Scanner coils: Do we need age-specific coils when dealing with a changing brain (e.g from infants to toddler to preschool)? How do we make sure that 'coil-fit' does not confound our analyses? 10/11

10) Multi-sites longitudinal studies: How do we keep the potential confounds as outlined above constant across sites/scanners? What guidelines do we need to establish? 11/11

Add on: 11) Points above 👆have implications for proposed sample size, attrition, power calculations. Guidelines in the field for these points will help to accurately estimate proposed sample sizes. Should these guidelines be different/less strict for studies in young children?