3dFWHMx and 3dClustSim on PET data

AFNI version info (afni -ver): Precompiled binary macos_13_ARM: Oct 29 2024 (Version AFNI_24.3.05 'Elagabalus')

Dear AFNI experts,

I am running 3dLME on PET data (cholinergic tracer) and would like to perform cluster correction on the results. I am using 3dFWHMx to estimate the autocorrelation in the data images (inputfiles for 3dLME) and use the '-2difMAD' option instead of -acf, because there is still much anatomical info in the data. After this I run 3dClustSim to find the minimum cluster size.
Does this approach sound right to you? Or am I missing something?
I get rather large cluster sizes out of 3dClustSim (~400 voxels).

Thanks!

Best
Hans van der Horn

As I am not a Federal employee, I will venture to answer. What you describe sounds reasonable - to me, who wrote the code you are talking about. However, I have never analyzed PET data myself with this or any method, so I cannot say how well it works for statistical control of errors. Testing would require a large collection of PET data, as has been done with FMRI data statistics.

Dear dr. Cox,

Thanks so much for your reply in the midst of this changing political landscape, I appreciate it. I agree that thorough testing is necessary before any definite conclusion can me drawn. We are open to any collaborations (we are clinicians with high affinity for neuroimaging methods) to dive further into this topic. For now, we will use the method lined out above to report our results.

Best
Hans