parcellation and functional connectivity analysis

[wicker]

I have performed a parcellation of several brains of two groups of subjects. My aim is now to use these parcells as anatomically informed regions of interest in a functional connectivity analysis of resting state fMRI data acquired in the same groups. I have used the 'create surface based functional data' with, as far as I know from the fact that the process didn't output any errors, success. I understand that this process should have extracted the time course at each nodes of my parcells, and for each subjects. I have two questions :
- Does a routine exists to average the extracted time courses of all nodes of a parcell ?
- What should be the best approach : average the signal in each parcell for each subject, build a correlation matrix for each subject and then average matrices in each group and then compare results ? or average the time course of each parcell in all individuals of each groups, build a correlation matrix for each group and then compare ?
- Would it make sense to also measure grey matter thickness in each parcell and study a potential correlation with (difference between groups in) functional connectivity measures ?
Thanks,

Regards

Bruno

[Olivier Coulon]

Hi Bruno,

I have used the 'create surface based functional data' with, as far as I know from the fact that the process didn't output any errors, success.

Well done, you're one of the first.

- Does a routine exists to average the extracted time courses of all nodes of a parcell ?

No but I will build a quick Python process to do that and send it directly to you.

- What should be the best approach : average the signal in each parcell for each subject, build a correlation matrix for each subject and then average matrices in each group and then compare results ? or average the time course of each parcell in all individuals of each groups, build a correlation matrix for each group and then compare ?

I would go for the first option, if only to be able to look at each subject's results. Also we have to think about how to perform the group-level stats, averaging the signal on a parcell and across subjects might be a bit too hardocre and won't take into account inter-subject (functional) variability. I think You will lose information if you do this. I don't really know the litterature on this topic...

Would it make sense to also measure grey matter thickness in each parcell and study a potential correlation with (difference between groups in) functional connectivity measures ?

yes it would, I guess. Does not hurt to have a look anyway. I'll look at how to get the grey matter thickness for each parcel. The main point is to transfer parcellation from the surface to the grey-matter volume mask, then the rest is easy. It's possible, I think there might even be a BV process for that, but I have to check it out. I'll come back to you on this...

Olivier