If you do:
D = spm_eeg_load;
D = sensors(D, 'EEG', );
That should fix it. I've seen this problem before and I thought I
fixed it, but I'll check again. Are you using the latest SPM update?
On 24 Mar 2011, at 17:45, "Tony W. Wilson" <[log in to unmask]> wrote:
> Dear SPMers,
> We are having a problem converting fif files that contain "false" EEG channels. Essentially, a few paradigms on the MEG acquisition computer are set to record EEG+MEG, but we do not always acquire the EEG because of time/subject/etc limitations. With these files, the EEG channels are sampled all same and included in the raw fif file, but the values are just noise. When we convert these using SPM8 (with MNE 2.7.0 toolbox), all appears well regardless of whether we import all the channels (stimulus, EEG, MEG) or just the MEG channels. However, there seems to be some sort of EEG flag that becomes embedded in the parameters file and creates problems later on. For example, in the case where we import all channels, we can declare all non-MEG channels as "Other" and then save in the "prepare" menu. However, once it is time to calculate a forward model, the 3D source localization gui still cues for an EEG model. If we select the BEM option and single shell (MEG), then we get the 'no good EEG channels' error message on the inversion - even though the gui reflects only MEG channels as available for this inversion. The same scenario happens when we import only the MEG channels during the conversion from fif. Furthermore, in the MEG-only import case, all the options for adjusting EEG electrodes remain in the "prepare" menus, which is not the case in true MEG only fiff files.
> In fif files without false EEG, we do not have any of these issues.
> I have searched the archives pretty well and have not seen a similar problem. We have seen this on Win-64 and CentOS 32-bit machines (both running current SPM8). Any thoughts on a fix for the false EEG files?
> Thanks in advance,
> Tony Wilson