I'm new to MRI analysis and don't think I can properly gauge the implications to subsequent spatial analysis of performing registration to the MNI152 template. The form of analysis I'm conducting involves segmenting ASD MRI scans by age to examine the disorder's developmental pathology, and thus an ideal study-specific method would involve generating a template for each age-range based on the scans from participants within that range (Ex: starting with 10 scans for each age range and 5 age ranges, generate a template for each age range from its 10 scans and use those for analysis). However, the difficulty of translating between study-specific coordinates and the more common MNI system when conducting future research is something I'd like to avoid if possible, so I want to understand the impact of registering all subjects to a single MNI template and performing this age-segmented analysis on these MNI-registered images. In either case I would create a template image for each age range- as I see it, the main difference would be whether or not these templates were registered to a standard coordinate system or not.
The subsequent analysis I'm planning to perform involves spatial correlation and the scans are dMRI-modal with plans to perform probabilistic tractography: I'm not sure to what extent transforming my database's scans would alter the results and was hoping for a more experienced perspective (or if my entire pipeline is flawed I'd greatly appreciate that heads-up as well!). Other papers such as this TMS analysis identifying structural asymmetry in ASD patients reports their data being "spatially normalized to the standard space," which makes it seem like this is typical for such analysis. Is this the case, or should I derive study-specific templates for my purposes?
