Back to the main page.

Bug 1599 - fixedori has no influence when using LCMV

Reported 2012-07-05 16:50:00 +0200
Modified 2012-08-23 14:02:12 +0200
Product: FieldTrip
Component: core
Version: unspecified
Hardware: PC
Operating System: Windows
Importance: P3 normal
Assigned to: Jörn M. Horschig
Depends on:
See also:

Jörn M. Horschig - 2012-07-05 16:50:18 +0200

next one :) fixedori is set correctly in beamformer_lcmv, the only problem is that sourceanalysis seems to not bother about this (or, as JM would probably say: it's a very stubborn beast!) %% draft of example script % load your very favourite data data = load(...); vol = load(...); % set basic cfg options cfg = []; cfg.vol = vol; cfg.method = 'lcmv'; cfg.fixedori ='yes'; oriyes = ft_sourceanalysis(cfg, data); cfg.fixedori = 'no'; orino = ft_sourceanalysis(cfg, data); % cfg is of course different, see above, so remove it orino = rmfield(orino, 'cfg'); oriyes = rmfield(oriyes, 'cfg'); isequal(oriyes, orino); % %ans = 1 %

Jan-Mathijs Schoffelen - 2012-07-09 10:59:27 +0200

Hi Jörn, Please retry with cfg.lcmv.fixedori = 'yes', rather than cfg.fixedori = 'yes'

Jörn M. Horschig - 2012-07-09 13:13:48 +0200

oh right, I forgot to mention, it only happens when using a pre-computed filter See also ft_sourceanalysis line 808 and onwards. tmpdip is computed using beamformer_lcmv (which is perfectly fine), but then dip is constructed without looking for ori (it does not matter whether it's cfg.lcmv.fixedori or cfg.fixedori).

Jörn M. Horschig - 2012-07-25 11:09:26 +0200

aha, I assume ori was not added, because it cannot be estimated for individual trials, however for now, I add the ori field to all single trial estimates. I also made adding a precomputed filter possible when the grid is not predefined (given that dimensions coincide). svn ci ft_sourceanalysis.m test/test_bug1599.m -m "bugfix-#1599-fixedori for lcmv works now also when using a predefined filter" Sending ft_sourceanalysis.m Adding test/test_bug1599.m Transmitting file data ..

Jörn M. Horschig - 2012-07-25 11:39:54 +0200

uh, I just saw this in ft_sourcedescriptives: % The following option only applies to LCMV single-trial timecourses. % cfg.fixedori = 'within_trials' or 'over_trials' (default = 'over_trials') within trials cannot be supported when using a predefined filter, given that all data will be concatenated for more efficient computation and ori will be computed on the concatenated data. something we need to hint to?

Jörn M. Horschig - 2012-08-23 14:02:12 +0200

bug closing time (