External Email - Use Caution
Dear Freesurfer experts,
I would like to know whether mri_glmfit can be used to estimate first-level (single subject) models as well or whether it is clearly not intended/recommended for that.
I have noticed that it allows to specify a design matrix explicitly and I used the one previously configured in SPM, estimated the individual subject models on the surface and got reasonable results. Then after reading a bit more I gathered that this function is usually used for second-level analyses, and that my models are lacking the temporal autocorrelation correction and do not take into account additional drift terms as one would usually do. I noticed however that there is as well the possibility to compute and save the temporal AR1 with the --tar1 option, but I do not understand how this is taken into account afterwards.
Would I get a proper first-level model by activating the --tar1 option and potentially adding additional drift terms to my design matrix myself, or would there be still remaining issues and the models should better be redone completely with selxavg3-sess (which seems a bit more complicated because the data were not organized in the required way from the start)?
Thanks a lot in advance,
Evelyn
It is just a GLM fitter, so it will take any data. Is this fMRI? Why not use a specific tool for that?
On 06/27/2018 08:51 AM, Evelyn Eger wrote:
Dear Freesurfer experts,
I would like to know whether mri_glmfit can be used to estimate first-level (single subject) models as well or whether it is clearly not intended/recommended for that.
I have noticed that it allows to specify a design matrix explicitly and I used the one previously configured in SPM, estimated the individual subject models on the surface and got reasonable results. Then after reading a bit more I gathered that this function is usually used for second-level analyses, and that my models are lacking the temporal autocorrelation correction and do not take into account additional drift terms as one would usually do. I noticed however that there is as well the possibility to compute and save the temporal AR1 with the --tar1 option, but I do not understand how this is taken into account afterwards.
Would I get a proper first-level model by activating the --tar1 option and potentially adding additional drift terms to my design matrix myself, or would there be still remaining issues and the models should better be redone completely with selxavg3-sess (which seems a bit more complicated because the data were not organized in the required way from the start)?
Thanks a lot in advance,
Evelyn
-- Dr Evelyn Eger INSERM Cognitive Neuroimaging Unit Neurospin - CEA Saclay F-91191 GIF/YVETTE, FRANCE
Tél : 33 (0)1 69 08 19 06 Fax : 33 (0)1 69 08 79 73
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Douglas N. Greve -
Evelyn Eger