[FieldTrip] Source activation vs baseline on MNE output

Rosy Southwell rosemary.southwell.14 at ucl.ac.uk
Mon Jan 22 21:50:38 CET 2018


Dear Fieldtrip Community,

I am working on EEG source analysis for auditory evoked responses to long
(3-second) stimuli. My data are evoked responses baseline corrected in the
interval [-1 0], and am interested in source activity in the window [0.5
1.5] relative to stimulus onset.

I have used MNE to estimate source activity over a latency of [-1 1.5]; see
code [1] below. Although I have two conditions of interest which I will
later contrast using ft_statfun_depsamplesT, I am first interested in
seeing which areas are activated by each condition. In order to extract
sound-evoked activity from ongoing activity, I would like to visualise the
source activity for each condition as an activation relative to baseline,
expressed as a T-statistic. I have ensured that my baseline window and
activation window are of equal duration and non-overlapping.

I have attempted to use ft_statfun_actvsblT for computing this statistic
(see code [2] below), but receive the error "Inappropriate dimord for the
statistics function FT_STATFUN_ACTVSBLT." From reading the function, I
understand that this method requires time-frequency data with
dimord 'chan_freq_time'. However my data is evoked, time-domain only.

My questions are
a) Is it even appropriate to express such "time-dimension-only" source
results as an activation relative to baseline?

b) If so, how would I best compute this in Fieldtrip?

c) if not, do you have a recommendation of how to quantify the degree of
activation for a single condition?

All the best,

Rosy Southwell
PhD Candidate
Chait Lab
Ear Institute, University College London.

%% CODE
% [1.] source analysis
  cfg = [];
            cfg.method = 'mne';
            cfg.latency = [-1 1.5];
            cfg.elec = elec;
            cfg.grid = sourcemodel_cortex;
            cfg.headmodel = headmodel;
            cfg.mne.prewhiten = 'yes';
            cfg.mne.lambda    = 3;
            cfg.mne.scalesourcecov = 'yes';
            [source] = ft_sourceanalysis(cfg, data);

% [2.] T-statistic of activation vs baseline

cfg=[];
cfg.method      = 'analytic';
cfg.statistic   = 'ft_statfun_actvsblT';
cfg.parameter   = 'pow';
cfg.correctm    = 'no';
cfg.alpha       = 0.025;
cfg.tail        = 0;
nsubj=20;
cfg.design(1,:) = [ones(1,nsubj) 2*ones(1,nsubj)];
cfg.design(2,:) = [1:nsubj 1:nsubj ];
cfg.ivar        = 1; % row of design matrix that contains independent
variable (the conditions)
cfg.uvar        = 2; % row of design matrix that contains unit variable (in
this case: subjects)

stat_RvR = ft_sourcestatistics(cfg,source_ac_all{:},source_bl_all{:});
% where source_ac_all, source_bl_all are 1*20 cell array of structs for
each subject
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20180122/a33ac592/attachment-0001.html>


More information about the fieldtrip mailing list