#1 Feature Request: average images preserving skull

To develop atlas-based brain extraction based on this template, it would be beneficial to also calculate T1w and T2w averages of the full head (i.e., before brain extraction).

Since the images used to create this template are not available to the public anymore, if the original subject-to-template transforms were kept, it would be fairly easy to just apply the transforms to the original images.

Thanks for the suggestion, Oscar. One issue I would foresee with creating average images from the original non-skull-stripped images is that the transformations were computed using brain extracted images. The deformation of structures outside the brain mask, in particular including the skull, may thus not be as plausible/smooth as desired. For example, nothing prevented the registration to "thicken" the skull when mapping the image into the common atlas space.

That said, I still have a copy of all transformations and should be able to give it a try. Please remind me to post an update if haven't!

Hi @schuhschuh, I wanted to follow up on this.

I agree that the deformation will not be great for the nonbrain areas. But indeed that is actually great - because the agreement will be lower, you will get some sort of smoothed average skull/face.

That is very helpful in atlas-based brain extraction since you need a registration algorithm to work but you lack in having precise brain masks.