Is it possible to keep track of trial-specific information in my fieldtrip analysis pipeline?

Sometimes it may be useful to keep trial-specific information tightly linked to your data structure. This trial-specific information for example can relate to the experimental condition, or to behavioural parameters.

Before July 2010 this was implemented in a rather obscure and undocumented way. It was possible to encode trial-specific information in additional columns to the trl-matrix, generated by a custom-written trialfun. The trl-matrix was passed on from one analysis step to another in the cfg structure, which in its turn was appended to the output data-structure. Subsequently, the trl-matrix was accessible by using the following command:

 trl = findcfg(data.cfg, 'trl'); 

This procedure was not only obscure, as mentioned, but also not consistent with the notion that a FieldTrip data structure should be self contained, and with the notion that the cfg-field in a FieldTrip data structure just serves the purpose to keep track of previous analysis steps. Therefore we extended the FieldTrip data structures with two fields:

  • the sampleinfo field, [Ntrials x 2] corresponding to the first 2 columns of the original trl-matrix
  • the trialinfo field, [Ntrials x M] corresponding to the optional 4th and upwards columns of the original trl-matrix.

Note that the 3rd column of the original trl-matrix is not represented in a separate field. In raw data (data structure containing trials) the offset is represented in each trial's time axis. A few things can be mentioned about the sampleinfo/trialinfo fields:

  • the sampleinfo field is only meaningful if the data structure is of datatype raw, and consists of epochs which were acquired in a single file, and with a sampling rate equal to the original digitization rate.
  • the trialinfo field is an optional field, which the user can manually add to a data structure. Alternatively, when a custom-written trialfun outputs a trl-matrix with more than 3 columns, the subsequent call to ft_preprocessing will automatically create a data structure with a trialinfo (and sampleinfo) field.
  • old-style (pre July 2010) FieldTrip data structures without sampleinfo/trialinfo can be created by
 data = checkdata(data, 'hastrialdef', 'yes'); 

Obviously, only a sampleinfo field will be appended if the data is a raw datatype and a trialinfo will only be appended if the original trl matrix contains more than 3 columns.

faq/is_it_possible_to_keep_track_of_trial-specific_information_in_my_fieldtrip_analysis_pipeline.txt · Last modified: 2014/04/15 15:22 by robert

You are here: startfaqis_it_possible_to_keep_track_of_trial-specific_information_in_my_fieldtrip_analysis_pipeline
CC Attribution-Share Alike 3.0 Unported
www.chimeric.de 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