"Grant, Steven (NIDA)" wrote:
>
> I am uncertain about the difference and interpretation of a conjunction
> analysis compared to masking of contrasts. For example, is inclusive
> masking of one contrast by another the same as a conjunction between the two
> contrasts ? When is it best to use conjunctions and when is it best to use
> masking ?
I'd be interested in seeing some sort of answer to this question, too.
My understanding from reading Cathy Price's paper (Price CJ & Friston
KJ, Neuroimage 5:261-270 (1996)) is that conjunction analysis is not the
same as masking one contrast by another, and that a conjunction analysis
between two contrasts is done as follows:
(1) Include as a regressor in the GLM the
interaction effect between the two contrasts.
(2) Threshold the Z map for the interaction
effect, generating a mask.
(3) Use this mask to mask the Z maps for the two
contrasts.
Is this right? Or, have there been new developments (e.g., the paper by
Worsley & Friston) which render the above obsolete? Has conjunction
analysis changed between SPM96 and SPM99?
Thanks,
Joe.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|