Diff for "SpmDemo" - Meg Wiki
location: Diff for "SpmDemo"
Differences between revisions 35 and 36
Revision 35 as of 2009-04-16 12:20:07
Size: 3625
Editor: RikHenson
Comment:
Revision 36 as of 2013-03-08 10:02:25
Size: 3629
Editor: localhost
Comment: converted to 1.6 markup
Deletions are marked like this. Additions are marked like this.
Line 23: Line 23:
'''Ok, no more waffle: here it is attachment:cbu_meeg_spm5_pipeline.m''' '''Ok, no more waffle: here it is [[attachment:cbu_meeg_spm5_pipeline.m]]'''

SPM5 CBU MEG (Neuromag) Demo

The previous single-subject demo (which included GUI instructions and a batch file) is here: SpmDemoOld. However, many improvements have since been made, so we now demonstrate analysis of a group of 8 subjects with concurrent EEG data.

Introduction

This demo takes MEG, EEG and MRI data from 8 subjects from the raw data to final group-level statistics in a template (MNI) space of differences between conditions in source strength during a time-frequency window of interest. The data were recorded at the CBU on our Neuromag MEG and Trio MRI machines, and are picked up directly from our /megdata/cbu/... and /mridata/cbu... network directories (as FIFF and DICOM formats respectively. For users outside the CBU, please email RikHenson for access to the raw data.

The subjects saw 172 trials of either faces or scrambled faces (the same paradigm used in fMRI, EEG and MEG by Henson et al, 2003, Cerebral Cortex; 2007, Neuroimage; in press, Neuroimage). Greater evoked energy for faces than scrambled faces is expected around 170ms (M170) from sources in bilateral fusiform and possibly right lateral temporal regions.

The demo is designed to illustrate some key unique advantages of SPM, eg use of SensorSpm to localise condition effects in (a 2D sensor) space and in time, data fusion (simultaneous inversion of MEG and EEG data) and group inversion. It generally follows this pipeline: BasicMeegPipelineSpm5

Notes on batch script

At the moment, we provide just a linear Matlab batch script - there are no accompanying GUI instructions yet (eg if you prefer pressing lots of buttons). Though the batch script does require some basic knowledge of Matlab (eg, arrays, cell arrays and simple "for..." and "if..." statements), we strongly recommend batching, for example if you have many subjects, want to keep a record of what you have done, or ever want to re-run with a new parameter/modification. In the future, we expect batching will be made easier (and more modular) by using AA http://imaging.mrc-cbu.cam.ac.uk/imaging/AutomaticAnalysisIntroduction

The script could clearly be improved (eg more comments, more modular, etc), but has been rushed out to meet repeated requests of users. It was deliberately made as one long file, to aid exposition, and it only has limited start-stop running capabilities. Its main purpose is to serve as the basis for your own batch scripts. Furthermore, it may change over the next few weeks, as people extend it (e.g, to include Jason's ICA or Olaf's channel surfing), or as people report problems. So feel free to improve and re-post (ensuring that functionality is maintained). Finally, note that a few of the options (eg group fusion of data) are not yet published (only submitted), and some parameters may be optimised in future (please do feedback with any discoveries about parameter choices, etc.)

To run the script below, you simply need to download it to somewhere on your Matlab path, create a new directory (eg /imaging/me01/CBU_SPM_MEG_Demo, and replace the line near the start of your version of the script with wd = '/imaging/me01/CBU_SPM_MEG_Demo'. The type spm 5 eeg at the Linux prompt, and in the new Matlab window, type the name of the script (or open the script in an Editor, and run it bit by bit by cutting and pasting or using the Matlab JVM debugger).

!!!Note: Script below was updated on 13.20 on 16/4/09 (added MEG noise estimates, PPMs, fMRI priors, and cleaned a tad)!!!

Ok, no more waffle: here it is cbu_meeg_spm5_pipeline.m

CbuMeg: SpmDemo (last edited 2013-03-08 10:02:25 by localhost)