Is mtnormalise necesary for single-subject analysis?

1

Hello everyone,

I am working at a project where we observe criticality of subject specific connectomes. It is clinical data with 1 b-value (1000) so i used SS3T with dwibiascorrect and dwi2fod dhollander but I did not mtnormalise. I did this because I will not compare connectivity matrices themselves nor graph metrics but rather a theoretical property for each subject, and at most compare individually at each level. But I’ve seen moreoften in this forum that mtnormalise is recommended. Is it reasonable to expect an invidualized response function better capture the subject specifc connectome, or will it just make a sloppier?

Best,
Ivan

2

Hi @Ivan_Mindlin,

My thoughts on this (I’m not an expert on this, so you might want to double-check): if you want to calculate some properties of the subjects not related with the CSD model, you should be fine without running mtnormalise, for example the mean FA in certain tracts. The FA isn’t affected by the usage of mtnormalise.

Having said this, indirectly the delineation of tracts will be affected by the usage of mtnormalise, changing the magnitude of the FODs, making some of them above or below the thresholds for tckgen.

So I would say that if you are planning individual analysis like the mean FA or something like this, you should be fine without mtnormalise and even without using the same response function to all the subjects, but any comparison you want to do using the CSD mode, you need to use it (and the same response function). I hope this helps.

Best regards,

Manuel

3

Hello Manuel,

Thank you for your answer!
In short, yes I will use the CSD model because I am constructing whole-brain structural connectivity matrices.
Regarding the same response function, would that mean in the case of running
mtnormalise wmfod.mif wmfod_norm.mif gmfod.mif gmfod_norm.mif csffod.mif csffod_norm.mif -mask mask.mif
I should store the first xx_norm.mif and use the same xx_norm.mif for all other patients?

Best wishes,
Ivan

4

Hi,

To use the same response function, once you calculate the responses for each subject, you have to average them using responsemean. Then use the averaged response functions to calculate each individual FODs, and finally, apply mtnormalise. From here, you can generate yor tractography, connectome, etc…

Best regards,

Manuel