Hello-
> Since I'm usually running very straight forward spm analysis this seems quite
> overwhelming to me.
If the design matrix itself contains (region-mean) BOLD signals I think
it is a bit less straightforward than usual. After putting those in it
sounds like a standard GLM estimate.
> What would be the proper methodoly for me to use to analyze the resting state
> data set I have? Is it under basic models > simple regression(Correlation)>
> where I would run one ROI time series per analysis . It seems the data in the
If you use one time series per design matrix that would remove the
problem of the non-orthogonal design. But then the next problem is how
the six maps (3 task positive, 3 task negative) relate to each other.
Not sure whether separate analyses were used in the paper though.
Usually in connectivity analyses, region time series are compared with
region time series, not voxel time series (using partial correlation or
something similar). Inference-wise, what does it mean if a voxel is
correlated with another voxel, but not with a task? It may lead to other
interesting analyses, but interpreting only that is quite tricky.
(see
http://cercor.oxfordjournals.org/cgi/content/abstract/15/9/1332
[http://www-bmu.psychiatry.cam.ac.uk/sitewide/publications/journal/salvador05neu.pdf])
>>each 110 volumes TR3sec
> i eyes closed
> ii crosshairs fixation
> Fox paper (in the figures) show each ROI as a seperate activation map.
>>each 110 volumes TR3sec
> i eyes closed
> ii crosshairs fixation
> iii no crosshairs
It is hard to reproduce the experiment + analysis from the paper (eg,
which conservative conjunction analysis was used?). Best is probably to
go straight to the source and ask the authors.
>>>activity). The degree of freedom was computed as n-3 so I guess there
>>>was no intercept (ie the data should already have zero mean).
>>Well, if I'm right (IF) the correlation analysis was performed on
>>residuals .. and thus assuming mean 0
Ah! They were residuals from a functional study. I thought the analysis
was performed on the resting state data themselves (after preprocessing).
>> Thanks for you time,
>> Maria
> No pblm, nice paper
> Good luck
> Cyril
Nice pictures
Best
Alle Meije
___________________________________________________________
Yahoo! Messenger - NEW crystal clear PC to PC calling worldwide with voicemail http://uk.messenger.yahoo.com
|