Nilearn image.smooth_img function

Hi Everyone,

A question regarding the image.smooth_img function from Nilearn, it smoothes the images by applying a Gaussian filter. I have been wondering if this function should be used on files when creating correlation matrices and seed-to-voxel maps?

Any help is appreciated!

Regards
Kelan

1 Like

Hi Kelan- nice to see you here =)
We should add that the niftiimage files we are smoothing have already been smoothed once via fMRIprep. So we are heavily leaning towards no- we shouldn’t double-smooth. But we thought it would be good to confirm this with the community.

As a general rule, I would not store smoothed data, but simply smooth them during the analysis. This is why the NiftiMasker object have a smoothing_fwhm parameter.
My 2c,