MEG Freeware Packages

This is a list of freeware packages more frequently used at the CBU:

SPM

Advantages:

• Much experience at CBU, fully integrated with fMRI analysis
• Unique strength in image statistics, eg Random Field Theory correction for Sensor-Time-Frequency images
• Fully flexible GLM, for complex designs (e.g, parametric modulation of evoked responses per trial)
• Novel variants of the L2-norm approach to the inverse problem (eg group inversion, EEG+MEG+fMRI fusion, canonical meshes in MNI space)
• Matlab, so readable and extendable

Disadvantages:

• Slowish graphics
• Limited head modelling (uses inverse-normalised template meshes)

Note there are two versions: SPM5 and SPM8. SPM8 is better in many ways, and should be used by people new to SPM. Prior development at the CBU with SPM5 however means that there is more support in terms of batch scripts etc. We hope these will be updated for SPM8 over the next year. The main advantages of SPM8 over SPM5 are:

• Fully compatible with Fieldtrip and EEGLAB (shared fileio, including the MNE Matlab functions to read FIF data; see below)
• Better GUI (fast enough to examine raw data)
• Better forward modelling (eg BEMs, using Fieldtrip code)
• DCM for evoked, induced and phase-coupling
• Bayesian dipole fitting

Official website: http://www.fil.ion.ucl.ac.uk/spm/

Currently installed on all Linux Boxes (lowest load box selected by spm 5 eeg or spm 8 eeg at linux prompt)

For more details, see here SpmAnalysis

MNE (Minimum Norm Estimates)

Advantages:

• Minimum Norm Estimation relying on minimal modeling assumptions, i.e. applicable to complex source configurations and high noise levels (by Matti Hämäläinen, a long-standing expert with strong links to Neuromag)
• Excellent, fast graphics

• Linked with Free Surfer http://imaging.mrc-cbu.cam.ac.uk/imaging/FreesurferInformation (advanced software for MRI segmentation, extraction of cortical surfaces, and inflation/normalisation)

• Matlab Toolbox for exporting and importing data (though this is independently available)
• Comprehensive documentation

Disadvantages:

• Compiled code, so difficult to extend (though it does come with an extensive MATLAB toolbox, which is arguably the best way to read FIF data).
• No statistics and time-frequency analysis (yet)

Official website: http://www.nmr.mgh.harvard.edu/martinos/userInfo/data/sofMNE.php

Installed on Linux Boxes. All Graphical tools must be used in the computing room on lws001, lws002, lws004, lws005, lws006 or using a vncserver.glx session (see http://imaging.mrc-cbu.cam.ac.uk/imaging/ImagingComputingDevelopments).

For more details, see here MneAnalysis

FieldTrip

Advantages:

• Strong in oscillatory analysis (time-frequency, coherence) and beamformers (eg DICS)
• In Matlab, so easily readable and extendable

Official website: http://www.ru.nl/fcdonders/fieldtrip/

Multiple Linear Regression of EEG/MEG data

Matlab tools for applying multiple linear regression to EEG/MEG data. More here: LinearRegression You may want to follow the following AnalysisPath

Data Screening Tool

Matlab utilities for quality control of your data prior to further processing. More here: DataScreeiningTool

Yet Others

• FIFFACCESS for simply reading FIF data into Matlab, more here: ReadingDataInMatlab

• EEGLAB (good for ICA) http://www.sccn.ucsd.edu/eeglab/

• Brainstorm (good for forward models) http://neuroimage.usc.edu/brainstorm/

• ...