Dear Paul,
I think this is one option, but a simpler alternative would be to do one analysis that includes both regressors (blood sample and cognition behaviour data), have a separate contrast for each and then do contrast masking between the two. These alternative approaches are only likely to be different in the case where there is a strong correlation between the two regressors, in which case a conjunction analysis is probably preferable.
All the best,
Mark
On 24 Jul 2013, at 03:19, Paul Chou <[log in to unmask]>
wrote:
> Dear all FSL experts
>
> Currently I want to investigate the relationship among blood sample data, brain image and cognition behavior data. I try to organize the analysis approach like follows, could some of you give me some advice about the study design ?
>
> Aim: To find the specific brain region which correlate the blood sample data and cognition behavior data simultaneously
>
> Step 1: Perform voxel wised multiple regression analysis between blood sample data and brain image data and set the statistical criteria as TFCE corrected P < 0.05
>
> Step 2: Perform voxel wised multiple regression analysis between cognition behavior data and brain image data and set the statistical criteria as TFCE corrected P < 0.05
>
> Step 3A: Using "AND" (Logic operation) analysis for result of Step 1 and Step 2. If there are some overlap area, I think these areas may correlate with the blood sample data and cognition behavior data simultaneously
>
> Step 3B: Using "Conjunction"analysis for result of Step 1 and Step 2 (Using the scripts from Thomas Nichols - http://www2.warwick.ac.uk/fac/sci/statistics/staff/academic-research/nichols/scripts/fsl/easythresh_conj.sh ) to find whether there are some brain regions correlate with the blood sample data and cognition behavior data simultaneously.
>
> Best
>
> Paul
|