External Email - Use Caution
Dear experts in longitudinal neuroimaging
We carefully read the papers about possible biases introduced by treating timepoints images differently (Reuter and Fischl NI 2011, for example) to estimate unbiased anatomical and functional changes of brain development in healthy individuals. We took good note that treating each time point independently would remove asymmetry biases at the cost of reduced power.
In our project, we would like to analyze about 800 scans both 3DT1, diffusion (FA preprocessed with fsl TBSS) and functional task BOLD at 3 time points (14 years-old, 19 years-old and 23 years-old).
Each time point scans are preprocessed individually and normalized to a specific group-based template at the corresponding time point (not a within-subject midpoint symmetric space) and to the standard MNI space.
Our question: We'd like to use a home-made study-specific multimodal atlas to resample ROI BOLD mean contrasts and fractional anisotropy at each time point but this atlas is *based on the last time point only (23 years-old).* Does this situation fall into the 'asymmetry-induced bias in longitudinal processing' ?
Our feeling is most probably yes but given the sample size this is likely to be marginal compared to the effect size and results would still be interpretable... ?
Many thanks for time, interest and insight,
Best wishes
Gabriel