Dear FSL experts,

Thanks much for your reply. I was waiting on this one! Our hypothesis is to see differences between SCA2 vs SCA1 and SCA3 patients. Earlier we had planned on comparing only SCA1 and SCA2 but now we have few more SCA3 patients so we have decided to include the SCA3 in our study too. I do want to see if VBM (gray matter volume) differs between SCA1+SCA3 and SCA2 patients.

I am trying to include CAG levels, do I have to demean it? If so in my GLM do I create two groups for comparison of SCA2 and SCA1-SCA3 or just 1 group with SCA1, SCA2 and SCA3?

Thanks and regards,

Sampada
AIIMS, New Delhi

On Mon, Jul 27, 2015 at 7:58 PM, Dr Sampada Sinha <[log in to unmask]> wrote:

---------- Forwarded message ----------
From: Anderson M. Winkler <[log in to unmask]>
Date: Mon, Jul 27, 2015 at 7:16 PM
Subject: Re: [FSL] Regarding: GLM design
To: [log in to unmask]


Hi Sampada,

There are various things that aren't clear on your description. Could you please give details of what is the hypothesis that you'd like to test? That is, do you want to see if VBM (gray matter volume) differs between SCA1 and SCA2?

About age and thickness: the fact that two or more nuisance variables aren't correlated with each other doesn't indicate that any of these should be dropped from the model.

About using a global measure of cortical thickness as a nuisance variable: I'm not sure this would be a good idea -- thickness and volume don't have a strong correlation, and even this relationship becomes irrelevant when the cortical volume is randomly split across voxels in VBM, and the spatial specificity of the thickness is lost through averaging. So, if there were a nuisance to be dropped, I'd consider thickness first, not age.

All the best,

Anderson


On 27 July 2015 at 10:32, Dr Sampada Sinha <[log in to unmask]> wrote:
Dear FSL experts,

Thanks for your reply. Appreciate all the valuable information you  provide to novice like us. In my case EV1 is SCA1 and EV2 is SCA2. Using the GLM gui and selecting the non-time-series design I got four contrasts. 

SCA1>SCA2
SCA2>SCA1
SCA1mean
SCA2mean


I have decided not to take age as a covariate and to take only cortical thickness, since looking at the preprocessed data I see no correlation of cortical thickness with age.

Thanks once again \1

Kind regards,

Sampada
AIIMS