Dear Varun,
This is a mixed design so always be careful with your contrasts and
prefer the simplest model using the summary statistics approach.
The contrast estimate plot shows you the difference time 1 - time 2
averaged across groups. In your case, have a look at the last F-contrast
in page 47 of the CAT12 manual, otherwise, if I'm not mistaken, try the
F-contrast eye(4)-ones(4)/4 to display the effect size for each of the
for levels of this 2x2 design.
Best regards,
Guillaume.
On 07/02/2022 20:49, Varun Arunachalam Chandran wrote:
> ⚠ Caution: External sender
>
>
> Hi fMRI experts,
>
> I have a doubt about defining a contrast matrix and a corresponding plot generated for a longitudinal event-related fMRI analysis. I used the CAT12 longitudinal analysis manual https://eur01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.neuro.uni-jena.de%2Fcat12%2FCAT12-Manual.pdf&data=04%7C01%7Cg.flandin%40UCL.AC.UK%7C9f2902f0977f47d05ef808d9ea7b60d6%7C1faf88fea9984c5b93c9210a11d9a5c2%7C0%7C0%7C637798638195597785%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=CWfwmhovjB%2B0zfCX7c85m1ac%2ByswSmWKbM6s5Rll%2FBw%3D&reserved=0 (page 43 and 47) as a reference for creating a design and contrast matrix for my analysis.
>
> One of the contrast of interest I used was, time 1 > time 2 in Groups A and B:
>
> 1 -1 1 -1
>
> I assume the contrast matrix compares the timepoints 1 and 2 in the patient group (A) and controls (B). In this case, a significant cluster was found at a predefined statistical threshold p<0.01. However, a plot generated corresponding to the result is not clear (Figure attached).
>
> The graphical representation (contrast estimate and 90% confidence interval) was expected to display two boxes indicating the comparison of time points patient group (A) and controls (B) in the figure. But unfortunately, the plots do not help very much to interpret the results. Please correct me if I am wrong? I would greatly appreciate your feedback/any materials for clarification.
>
> thanks,
>
> kind regards,
> Varun
>
--
Guillaume Flandin, PhD
Wellcome Centre for Human Neuroimaging
UCL Queen Square Institute of Neurology
London WC1N 3BG
|