Dear Darren and SPMers,
Sorry to bring up this question and repeat it again, but I am confused by the answers regarding the "VOI data adjust" issue on the list and the results I got.
What puzzles me is that, by adjusting for an F contrast, the time series is corrected by removing the effects INCLUDED in this F contrast or removing the effects EXCLUDED in this F contrast?
The previous answer (quoted bellow) clearly states that adjusting for an F contrast will correct the time series by removing the effects excluded in the F contrast (correct me if I am wrong).
However, the result I got seems to be the opposite. In my case, there are four conditions/stimulus types and I set a F contrast of effects of interests which is [1 0 0 0; 0 1 0 0; 0 0 1 0; 0 0 0 1] (righted padded with zeros; so includes the effects of interests and excludes the headmotion parameters and constant), and select this F contrast for the "adjust data for" option when creating a VOI (located at an activated area). The resultant 1st eigenvariate looks very noisy. But, if I define a new F contrast including the confounds and excluding the experimental conditions (i.e., zeros for the first four columns and ones for the six headmovement parameters and constant column), the resultant 1st eigenvariate looks much more reasonable to me - the time series change with the stimulation.
It seems to me that adjusting for an F contrast removed the effects INCLUDED in this F contrast. But obviously I am wrong according to the SPM manual and the previous answers on the list. Can anybody explain and clarify this?
By the way, I am using SPM8.
Best,
Meng
On Thu, 5 Nov 2009 12:23:05 -0600, Darren Gitelman <[log in to unmask]> wrote:
>yes.
>
>darren
>
>On Thu, Nov 5, 2009 at 9:07 AM, Anne Kathrin Rehme <[log in to unmask]>wrote:
>
>> Dear Darren,
>>
>> thank you for your answer. In detail and just to be sure that I understood
>> your mail and Klaas' mail correctly: By adjusting for an F contrast that
>> INCLUDES experimental regressors of interest and EXCLUDES regressors of no
>> interest, one would remove the effect of the regressors of no interest from
>> the VOI time series?
>>
>> Best regards,
>> Anne
>>
>> ----- Urspr�ngliche Nachricht -----
>> Von: Gmail <[log in to unmask]>
>> Datum: Donnerstag, 5. November 2009, 13:50
>> Betreff: Re: [SPM] How to adjust VOI time series for DCM?
>> An: [log in to unmask]
>>
>> > Anne
>> >
>> > This is confusing, but Klaas probably meant adjusting for confounds by
>> using an effects of interest contrast, so it's really speaking about the
>> same thing. (The note from Klaas also comments on adjusting for other things
>> which he advises against unless there is A good reason.)
>> >
>> > -----
>> Darren Gitelman, MD
>> > sent from my iPhone
>> >
>> > On Nov 5, 2009, at 4:09 AM, Anne Kathrin Rehme <[log in to unmask]>
>> wrote:
>> >
>>
>> > Dear SPM users,
>> >
>> > I wonder whether the VOI time series has to be adjusted for the effects
>> of interest contrast (which contains only the experimental regressors of
>> interest) or for the confound (i.e., F-contrast for the
>> > > regressors one is not interested in, e.g., the realignment
>> parameters). I know that this question has been discussed before but I am a
>> little bit confused because I found somehow inconsisting answers. In the
>> SPM8 manual on page 291, adjustment via effects of interest is recommended.
>> By contrast, in a recent email from Klaas Enno Stephan, he advises to adjust
>> the data for anything that is known as a confound, e.g., realignment
>> parameters (<https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind06&L=SPM&P=R789474%29.><https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind06&L=SPM&P=R789474>
>> https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind06&L=SPM&P=R789474).
>> >
>> > Can anyone explain to me in which case a particular adjustment has to be
>> made?
>> >
>> > Any advice is very much appreciated,
>> >
>> > Anne
>>
>>
>
|