SPM corrects for sphereicity in the data, using a sub-set of voxels which have a significant main effect based, if I recall correctly, an f-test run internally during the 1st level analysis.
The error is that there are no such voxels to test sphereicity against. This also implies a problem either with the model or the data. By problem, it could mean the data is not good, the model has an error and does not properly match the data (e.g. the event timings are wrong), or you may have run an experiment which just does not really evoke a BOLD change.
Every time I have seen this error in my on work, it has been a harbinger of problems and a total lack of effect. For example, in one instance, we had noisy data, and I believe we 'over-preprocessed' it, so there was not enough signal variance related to the task for form a decent model.
You can in principle reduce the statistical threshold to try to get around this. I am not sure I would recommend this. However in spm_defaults, it is defaults.stats.fmri.ufp
Good luck,
Colin Hawco, PhD
Neuranalysis Consulting
Neuroimaging analysis and consultation
www.neuranalysis.com
[log in to unmask]
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Kari Lehikoinen
Sent: September-20-16 12:48 PM
To: [log in to unmask]
Subject: [SPM] Problem with 1st level analysis in SPM; "no significant voxels"
Hello everyone
I've been trying to do 1st level analysis in SPM, and keep running into a "no significant voxels" -error. There is also another error message before that, which I've attached to this message.
I have several subjects and several conditions, and this problem only happens for some of the subjects, and only in one condition, despite my batch scripts being similar (also attached).
So, I guess my question is: is this real error message, or do those cases really just not have significant voxels? I've searched the archives, and although this error message seems common, I haven't found a situation that matches mine.
Thanks for anyone who can help.
- Kari Lehikoinen, University of Helsinki
|