Either you haven't digitized some of the electrodes or you have some channels whose type was wrongly set to EEG. I suggest that rather than creating an elec struct you make an .sfp file with your channel locations and fiducials  and load it with 'Prepare'. That will ensure that the struct is in the correct format and will work in both SPM8 and SPM12. If you intend to fuse your EEG with MEG you should load the sfp file after fusion. Also note that you should either convert your EEG data to average reference or if you use SPM12 you could specify the reference montage after loading the sensors.

Best,

Vladimir

On Tue, Sep 9, 2014 at 12:30 PM, Roni Tibon <[log in to unmask]> wrote:

Thanks Vladimir!

I'll look into it. 

I have another question that has to do with the sensors' locations – this time of EEG data, prior to fusion. 

I was digitizing the locations of the 62 EEG channels while I was recording the data, and would like to use these locations (instead of the default layout, which I was using before). 

I used the digitized data to create an elec struct, and then used D = sensors(D, 'EEG', elec), and spm_eeg_prep with project3D.

Again, everything looks good when I display the locations using display > M/EEG > EEG scalp. However, when I'm trying to use the 3D source reconstruction I get an error message saying that the requested file is not ready to use, and the matlab command window says “checkmeeg: not all EEG channel locations are specified”. 

What am I doing wrong?  

Thank you for your time, 

Roni

On 9 September 2014 12:13, Vladimir Litvak <[log in to unmask]> wrote:

Dear Roni,

I don't think it's a problem. You just have the complete extended 10-20 channel set in your sensor representation, but for source reconstruction etc. only the channels you actually have will be used. What is more worrying is that in the coregistration display your EEG sensors appear to be inside the head. This is not a huge problem as the locations will eventually be projected to the surface, but it might not be the most precise thing to do. I suggest you try reloading your EEG sensor locations after fusion. You will be asked to match between EEG and MEG fiducials and then all sensor locations will be put in the same space that should hopefully match your head model.

Best,

Vladimir

On Tue, Sep 9, 2014 at 8:06 AM, Roni Tibon <[log in to unmask]> wrote:

Dear SPM experts, 

I'm using spm_eeg_fuse to fuse EEG data (62 channels) with MEG data (248 channels). 

After fusion, for D.sensors('MEG') I get the 248 MEG channels. However, for D.sensors('EEG') I get 348 channels!

D.modality returns 'Multimodal'. D.sensors('EEG').label returns strange looking labels such as: 

'FFT9h'

'FFT7h'

'FFC5h'

'FFC3h'

'POO2h'

'POO4h'

'POO6h'

'POO8h'    

'AF7h'

'AF5h'

'AF3h'

'AF1h'

This is on top of the actual 62 EEG channels. 

I see these additional channels when I display the EEG channels after 3-D source reconstruction, but not when I display them using display > M/EEG > EEG scalp (see figures). 

Same results with SPM8 and SPM12b. 

What's going on here? 

Thanks for your help, 

Roni Tibon

--

   /\.../\
  ( -- ---)__{}
   (_.._...._)

http://shkafkafim.blogspot.com/

--

   /\.../\
  ( -- ---)__{}
   (_.._...._)

http://shkafkafim.blogspot.com/