# Deconvolution kernel across the whole brain? **URL:** https://community.mrtrix.org/t/deconvolution-kernel-across-the-whole-brain/1855 **Category:** Uncategorized **Created:** [August 22, 2018, 5:08pm UTC](https://community.mrtrix.org/t/deconvolution-kernel-across-the-whole-brain/1855 "2018-08-22T17:08:42Z") **Posts on this page:** 1 **Showing post:** 3
### Author: ![jdtournier](https://community.mrtrix.org/user_avatar/community.mrtrix.org/jdtournier/32/2594_2.png) [@jdtournier](https://community.mrtrix.org/u/jdtournier) #### Post date: [August 25, 2018, 2:02pm UTC](https://community.mrtrix.org/t/deconvolution-kernel-across-the-whole-brain/1855/3 "2018-08-25T14:02:42Z")
> [@zeydabadi](#): > > Any thoughts? Too many, unfortunately… > [@zeydabadi](#): > > Is it assumed in “[Fibre density and cross-section - Single-tissue CSD](https://mrtrix.readthedocs.io/en/latest/fixel_based_analysis/st_fibre_density_cross-section.html)” that the deconvolution kernel is the same across the whole brain? Yes. That’s a central assumption in any spherical deconvolution approach, whether in _MRtrix3_ or otherwise. That’s extended in multi-tissue CSD to a _set_ of kernels, each of which is assumed the same across the whole brain. > [@zeydabadi](#): > > If so, how can we extend this to the infants’ data? It depends what you mean by ‘infants’ exactly. If neonates, then you can use simple approaches as suggested in [this thread](https://community.mrtrix.org/t/multi-shell-data-in-neonates/561/3?u=jdtournier). See [this thread](https://community.mrtrix.org/t/neonatal-multi-shell-data-optimal-preprocessing/1238) for a more involved discussion of the difficulties in modelling such data (there’s a few others on the topic if you search around). Otherwise, there’s been a bit of work on this presented at the last ISMRM, but it’s still very much an open question. If on the other hand, by ‘infants’ you mean children over 1 or 2 years of age, then I reckon the standard adult pipeline would work just fine. The kernel is estimated from the data, so should match their particular properties – it might differ from what you’d see in adults, but probably not by a great deal by that age. If by ‘extend’ you meant how do we include infant and adult data together in the same analysis, then that’s a whole other question… But again, if ‘infant’ refers to subjects aged 1-2 or older (preferably at least 2), then you _might_ find that you can analyse them all using the same response function(s), as we recommend for any group analysis. But this is a gut feeling: the validity of this recommendation would need to be verified somehow, and furthermore will depend heavily on the specifics of your research question… --- _[View the full topic](https://community.mrtrix.org/t/deconvolution-kernel-across-the-whole-brain/1855)._