Dear Ulrike
Thank you for taking the time to report this. There were some issues with the computation and display of the DCM kernels, which caused the issue you reported. It shouldn't change your model comparisons or parameter estimates. The issue has been fixed for the next public release of SPM. If you need a copy sooner than that, please get in touch with me.
Best
Peter
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Ulrike Horn
Sent: 14 March 2016 10:36
To: [log in to unmask]
Subject: [SPM] Hemodynamic responses in DCM (SPM8 and 12)
Dear SPM users,
I have been experiencing some issues with different results in SPM8 and SPM12 and would like to ask what this is all about.
When getting some flat lines after estimation of my data with DCM in SPM8, I tried the newest SPM12 version after I read about this problem here.
This works better for estimating a non-zero solution on noisy data but I noticed some issues regarding the hemodynamic response estimation.
I added some pictures of the hemodynamic responses (stored in DCM.H1) from estimating the example data set for attention to motion. With SPM8 and the default configuration (as described in the manual) the responses are as expected whereas with SPM12 and default values the curves are often negative, converging at the beginning, or consist of smaller values. I also changed the TR in SPM12 to other values to match the SPM8 defaults but the result is the same.
Is it important how these responses look like? If I experience the same with my data - can I still trust the parameter values for the overall models?
Or have I done something wrong here?
Any help or advice would be really appreciated!
Thanks.
U. Horn
|