Smoothing pre ICA-AROMA - why 6mm?


I understand that fmriprep follows the protocol in this ICA-AROMA paper, where they use a 6mm smoothing kernel. However, here one of the authors of that paper says that the 6mm was chosen as ceil(1.5 * the voxelsize). Shouldn’t this then also change in fmriprep? presumably that’s a very easy change? In my data, for example, the voxel size is 2mm isotropic and I’d like to use the non-aggressive AROMA denoised data, but smoothing with a 6mm kernel seems too much…

Does this make sense?


Hi @alonbaram,
I did alot of reading on this about a year ago so forgive me if I am rusty. At the time I found this paper from Chen & Calhoun that discuss the effects of kernel size for FWHM smoothing:

They state that:

Our experimental results suggest a 2~3 voxel FWHM spatial smoothing for single-subject ICA in achieving an optimal balance of functional connectivity, and a wide range (2~5 voxels) of FWHM for multi-subject ICA.

So it seems like in your case that ICA AROMAs 6mm kernel would be along the right lines.

Best of luck,

1 Like

Great, thanks! that’s helpful to know.