Hello,
Thank you very much for your prompt reply.
We meant to use both, hippocampal thickness as well as volume measures. We are attempting to correlate fmri activation patterns ( new versus repeated faces) in 'young' vs 'elderly' healthy adults to subject's hippocampal volume and thickness values.
Would you suggest to go about this a different way?
cheers,
Catherine
________________________________________
From: MCLAREN, Donald [[log in to unmask]]
Sent: Friday, March 09, 2012 3:46 PM
To: Chong, Catherine - SJHMC
Cc: [log in to unmask]
Subject: Re: [SPM]
On Fri, Mar 9, 2012 at 2:44 PM, Chong, Catherine - SJHMC <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Dear Experts,
I am new to SPM, so please excuse my question.
I seek to evaluate a memory paradigm (new versus repeated faces) for a patient and a control group of participants. I would like to correlate mean cortical thickness values for the hippocampus into my contrast design, i.e. my question is: Is hippocampal thickness correlated to fmri brain activation patterns? Do I implement the mean cortical thickness values as a covariate in my analysis in a 2-sample T-Test?
(1) Are you using cortical thickness or hippocampal volume? If you are you using thickness, is there a reason that you chose thickness?
(2) As for the analysis, there are are couple of approaches.
(a) two-sample t-test with one covariate that is demeaned. This will give you the covariate-adjusted means. Assumes the slope is the same in both groups.
(b) two-sample t-test with one covariate. This will give you the group intercepts. Assumes the slope is the same in both groups.
(c) two-sample t-test with two covariates that is demeaned within-group. This will give you the group means controlling for the covariate. You can also test is the slopes are different.
(d) two-sample t-test with two covariates that is demeaned across everyone. This will give you the covariate-adjusted group means. You can also test is the slopes are different.
(e) two-sample t-test with two covariates that is not demeaned. This will give you the covariate-adjusted group intercepts [I believe]. You can also test is the slopes are different.
Models c-e all have the same slope estimates.
If there is an difference in the slopes, you should not interpret the group means/intercepts. The average of the slopes might also be difficult to interpret.
Thank you very much for your help,
cheers,
catherine
