Fmriprep: resampling to different MNI-space using template with or without skull?

1

Hi,

I want to specify an individual output space and I am asking myself if I need to provide a template including or excluding skull (and meninges). What is the optimal choice?

thanks,

Eddy

2

Hi,

so I preprocessed an Image and specified once a skullstripped image as desired output space and once an image including skull. As expected, the skullstripped image results in a smaller brain, the template with skull in a bigger one. Intuitively I´d prefer the skulltstripped as template, but I actually have a hard time judging this, because using the skullstripped seems to cover some meninges, which the other image does not.

What would you recommend?

When not specifying an individual output, what kind of MNI-template is fmriprep using for normalization and further coregistration: A skullstripped template?

thanks,

Eddy

3

Hi Eddy,

If you want to use a manually specified template, it needs to be in a particular format in order for us to correctly use it. This includes unmasked template files AND binary brain masks.

Best,
Chris

4

Hi,

thank you for this answer! I had to use an older version of fmriprep (for reasons of consistency), where I specified the --template-resampling-grid command . What would be the recommended option here: With or without skull?

I wonder if I better reprocess my datasets with a newer version.

Best,

Eddy

5

Ah, thank you for clarifying… My inclination is to say that the normalization is done from a skull-stripped brain to a skull-stripped template, and that therefore you will want the skull-stripped template.

@oesteban (on vacation, so may not be quick to respond) would be a better person for the details of how custom templates were previously specified. They were not very well supported for basically the reason that, to be used correctly, a template does need multiple files.

6

HI,

using the skullstripped brain was also my intuition, although the skullstripping on functional images often left some residuals which could be seen as the outer boundary of the skull and some meninges (and noise of course). This made me unsure about what to use.

And it is hard for me to judge from trying out. Both versions seem to have pros and cons. I´ll wait for the answer of @oesteban .

I will try out a new version of fmriprep and compare the output as well: This will tell me which version was supposed to be used.

Thanks,

Eddy