I don't think there are clear answers to this kind of question. The optimal smoothing kernel size depends on the size of the effect you are looking for, not the modality that you are using. The actual amount of smoothing will be more for PET because of the inherent smoothing, but that smoothing is 3D, not 2D, so it is hard to compare.

On 5/30/17 10:08 AM, Eduard Vilaplana Martinez wrote:

Dear FS experts,

I've been working with the proposed pipeline, but I still have a doubt regarding the degree of smooth to be applied to the final surfaces prior to the analyses.

The first question is that, given a "standard" PET scanner, with a PSF between 5-8mm, what size smooth would you apply? Would 10mm or 15mm be reasonable (as for MRI)?

The second question is if it is correct, in your opinion, to use the same size of smooth kernel for a multimodal PET and MRI analysis (i.e., local intermodality correlations, for example FDG and MRI)? Or, given the more smooth nature of the PET signal, would you apply a lower degree of smooth to the final PET surface?

Thanks in advance for your advices, Edu V.

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer