Getting started with EGI data

Please add information if you're using EGI data and want to share info with other FieldTrip users.

Introduction for the EGI file formats

Net Station can export data to several data formats that FieldTrip can read in.

  • *.egis, *.sbin, and *.mff can be read in by FieldTrip (I have no details on how well they work, and what the known issues are). Other formats like *.ave, *.gave, *.ses are not supported but the data can be read by exporting to one of the supported formats.
  • Simple binary (.raw), only the channels from the hdEEG net are exported, the PIB channels, for instance, are not. Channel labels are not present in this format, and are thus made on the fly by FieldTrip. Events are present and read in correctly. Data sets containing multiple epochs are exported by NetStation as separate .raw files. The events are, however, exported as one file, which make it tricky (read impossible) to align data and events properly when having multiple discontinuous epochs.
  • EDF+ (.edf) All channels, including PIB channels, are read in correctly, including channel labels. However, the events, which are stored on the annotation channel, are written in a way by Net Station that is not compatible with the edf+ reading implementation in FieldTrip. So, events do not come out properly. Also discontinuous epochs are “glued” together as one “continuous” data stream.

Meta File Format (mff)

The beta version of NetStation 4.5 (and up) writes data in mff, which is now (April 21, 2011) fully supported by FieldTrip. FieldTrip reads in the data of all signals (hdEEG-net, PIB, etc). Data sets containing multiple epochs are supported, FieldTrip keeps track of the offset to the start of the recording for all epochs. Also events are read in, and also with discontinuous epochs aligning of events to data works.

The initial (v1) implementation was made by Ingrid Nieuwenhuis. It works for quite a few datasets but also has a number of known limitations (a.o. not full reading of info from triggers/events).

The second (v2) implementation is provided by the EGI company and is based on their general-purpose Java library and should support all data format features. As soon as the V2 code works at least as good as the old V1 implementation, the 'egi_mff' case will move from the v1 to the v2 implementation, causing the default behaviour to switch.

In short, the code in ft_read_header and ft_read_data does

case {'egi_mff_v1' 'egi_mff'}
 % do the old stuff
case {'egi_mff_v2'}
 % do the new stuff

At this moment the default is to use the egi_mff_v1 implementation. This can be overruled by specifying

cfg.dataformat = 'egi_mff_v2'
cfg.headerformat = 'egi_mff_v2'

to ft_preprocessing and all other high-level FT functions that read data. Furthermore, by specifying

ft_read_header(...., 'headerformat', 'egi_mff_v2')
ft_read_data(...., 'headerformat', 'egi_mff_v2', 'dataformat', 'egi_mff_v2')

also the low-level fileio functions will use the v2 reading functions.

Installing the EGI Java implementation

The egi_mff_v2 implementation in MATLAB depends on a generic Java implementation of the reading functions.

To ensure that MATLAB can find the .jar file, you have to add a path to the MFF-1.0.jar (java) file by typing the following on your matlab prompt

javaaddpath path/to/fieldtrip/external/egi_mff/java/MFF-1.0.jar

For more information have a look at javaaddpath and javaclasspath.

getting_started/egi.txt · Last modified: 2016/04/28 13:58 by robert

You are here: startgetting_startedegi
CC Attribution-Share Alike 3.0 Unported Valid CSS Driven by DokuWiki do yourself a favour and use a real browser - get firefox!! Recent changes RSS feed Valid XHTML 1.0
This DokuWiki features an Anymorphic Webdesign theme, customised by Eelke Spaak and Stephen Whitmarsh.
Mobile Analytics Website Security Test