Dear Peter,
Thank you very much for your quick answer!
As you presumed the outputs of my models look like a flat line in each
region (see attached file). I thought this would be concordant with the
variance of the extracted timeseries of 0%.
Best,
Eric
> Hi Eric,
> There are two separate things here to consider. One is how much of the variance in your region is explained by your timeseries (i.e. the PCA which happens when you press the "eigenvariate" button in SPM). That's checked using the code you mention below. The second thing is how much of the variance of the extracted timeseries is explained by the DCM, which you found is 0%. This suggests the model hasn't worked. If you click the DCM button in SPM, then Review, then Outputs, does the model look like a flat line in each region?
>
> (One thing to rule out is that spm_dcm_fmri_check doesn't have a problem with large models - this is unlikely, but let us know about the above first.)
>
> Best,
> Peter
>
> Peter Zeidman, PhD
> Methods Group
> Wellcome Trust Centre for Neuroimaging
> 12 Queen Square
> London WC1N 3BG
> [log in to unmask]
>
> -----Original Message-----
> From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Eric Holst
> Sent: 19 November 2014 17:35
> To: [log in to unmask]
> Subject: [SPM] DCM: Problem with percentage of explained variance
>
> Dear DCM experts,
>
> for an event-related fMRI study (16 healthy participants with 6 sessions each) I want to run a DCM analysis to investigate attentional modulation between regions V1, V4 and PPC of both hemispheres. I am currently using SPM8, release 4290.
>
> So far I have specified and estimated one GLM for each participant. For VOI extraction I have defined 6 ROIs for V1, V4, PPC of both left and right hemisphere (masked inclusively with SPM Anatomy Toolbox, threshold p<0.05 uncorrected, global maximum, sphere 6 mm), resulting in 36 VOIs per participant. I have checked the number of voxels in the VOIs (mostly > 30) and - as suggested in an older post - I have checked for the percentage of signal’s variance in the VOIs, which is always >65%.
>
> Matlab code:
> load VOI_xxx.mat;
> 100*xY.s(1)/sum(xY.s)
>
> The basic setup of each DCM is similar to the one used for the „attention to motion“ paper, including 6 regions with reciprocal intrahemispheric connections between V1-V4 and V4-PPC and reciprocal interhemispheric connections between V4-V4 and PPC-PPC. Driving input is allocated to both V1 regions. On this basis I have set up multiple DCMs with alternating modulatory input to either regions or connections to subsequently perform a BMS analysis.
>
> So here is my problem:
> When I run „spm_dcm_fmri_check“ on any of the estimated DCMs, I get an „explained percentage of variance“ of 0%! How is that possible, given that the percentage of signal’s variance of each VOI is > 65%?
>
> In older posts I have read about the problem of „flat lining“ with SPM8, but the same problem occurs when using SPM12 for specification and estimation of the DCMs (I haven’t re-run the GLM).
>
> Any help is greatly appreciated and thank you very much in advance.
>
> ---
> Eric Holst
> Department of Neurology
> Charité, Berlin, Germany
|