0
$\begingroup$

I'm having an issue with deciding how to correct for multiple comparisons with my structural MRI region of interest analyses. I have two groups, Group A and Group B, and am looking at cortical thickness differences between the two groups.

Before I ran the analyses, I chose 4 regions of interest to look at. I created the cortical thickness maps and then conducted individual ROI analyses for each of the 4 regions, using FSL Randomise.

FSL Randomise does correct for multiple comparisons, and did so for each individual region. However, this doesn't take into account that I ran 4 ROI analyses. So I was wondering how I should account for each ROI analysis.

Thank you for your time!

$\endgroup$
0
$\begingroup$

A few comments, There is quite a bit of jargon in your post. For example, even if someone was working in fMRI they might not use FSL. It helps a little bit to give some explanation before jumping straight in.

Secondly, I have one question, if you are performing 4 ROI's on a participant's output, why not just use the threshold map and be done with it? Particularly, why not just use FSL's higher order analysis to look at the mean threshold map. That has where the clusters are located already in it. Unless you have a specific reason to do this based on your hypothesis, I am just not certain what the particular benefit might be.

$\endgroup$
  • $\begingroup$ Thank you for your suggestion, apologies for the jargon! The reason I'm using FSL Randomise and not FSL FEAT, is because I am using structural data and not fMRI data. I find Randomise easier to use with structural data. $\endgroup$ – Guest_user Oct 23 '17 at 14:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.