Dear Kambiz,
I also use an Elekta Neuromag system and I have been able in the past to convert MNE-pre-processed (i.e. mne_process_raw averaged, artifact-removed) data from *.fif format to spm8 format in a straightforward manner, using the convert button in the GUI, or the basic conversion matlab scripts. The fiducials and sensor positions were maintained fine and source reconstruction was possible.
Of course, if you import trial-averaged data, your subsequent analysis in spm8 will necessarily be limited to average-based analyses: i.e. you won't be able to utilize trial-wise source reconstructions/statistics utilities!
Best wishes,
Nela
----
Nela Cicmil
D.Phil Candidate, Neurophysiology
Dept. Physiology, Anatomy & Genetics
University of Oxford
Tel: 01865 282274
Internal: 82274
________________________________________
From: SPM (Statistical Parametric Mapping) [[log in to unmask]] on behalf of Vladimir Litvak [[log in to unmask]]
Sent: 13 March 2012 21:14
To: [log in to unmask]
Subject: Re: [SPM] MEEG artifact rejection
Dear Kambiz,
On Tue, Mar 13, 2012 at 6:49 PM, Kambiz Tavabi <[log in to unmask]> wrote:
> Hello all,
>
> I'd like to use SPM 8 to carry out exploratory group level statistics on MEG data collected on a 306 channel Elekta system. I have a couple of questions; (1) I've preprocessed my data according to the SPM 8 manual up to the point of artifact rejection. I would like to use the peak-to-peak algorithm on the planar gradiometers only with a threshold value of 3000fT/cm. When I set up a batch process to implement the artifact detection module with values 3000 or 3e-10 I get discrepant and/or worrying results, i.e., rejecting nothing or every trial in the dataset. I assume I am not entering the correct value translating to 3000ft/cm. As a side note, I get different results with the MNE software peak-to-peak algorithm on the planar gradiometers (>3000ft/cm), albeit with different filters and sampling rates between the software. Any ideas?
I think the data read to SPM is in T/m but I might be mistaken. The
units should be displayed in the reviewing tool in info/channels. The
best way top set the threshold is plot the data and see what values
make sense. If you do that I think you'll quickly see what the right
order of magnitude is.
(2) Is it possible to carry out 2-nd order sensor level statistics on
preprocessed, trial-averaged data (e.g., my MNE pre-processed fif
data) after converting to SPM and creating images?
>
Yes, that's perfectly OK. I'm not sure how MNE handles the data but if
it generates fif files containing valid sensor and fiducial
description you should also be able to do the other things - 3D source
reconstruction and DCM. If not, you might need to re-read the sensors
and the fiducials from the original fif file using spm_eeg_copygrad.
Best,
Vladimir
> Thanks in advance
> Kambiz
|