Back to the main page.
Bug 718 - ft_apply_montage of ft_megplanar causes ft_topoplotTFR problems after ft_rejectcomponent
| Status | CLOSED FIXED |
| Reported | 2011-05-31 17:48:00 +0200 |
| Modified | 2011-07-13 14:22:22 +0200 |
| Product: | FieldTrip |
| Component: | core |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Linux |
| Importance: | P1 normal |
| Assigned to: | Jan-Mathijs Schoffelen |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: |
Diego Lozano Soldevilla - 2011-05-31 17:48:00 +0200
After rejection of ICs by means of ft_rejectcomponent, I get topoplotTFR display problems (it appears the electrodes and the topo scalp but without frequency information). Importantly is that this problem doesn't appear with ft_multiplotTFR. Lines 112-121 of ft_apply_montage seems to cause the problem. Specifically, without rejecting ICs, the sens.label and montage.labelorg have different channels and this extra channels (28) seem to be important for topoploting. Rejecting ICs, this extra channels not appear in megplanar calculations and latter appear the topoplot problems.
Diego Lozano Soldevilla - 2011-05-31 17:48:35 +0200
Created attachment 60 data example
Jan-Mathijs Schoffelen - 2011-06-24 15:52:05 +0200
Created attachment 77 data causing problems in channelposition
Jan-Mathijs Schoffelen - 2011-06-24 15:59:56 +0200
Created attachment 78 yet other data causing problems in channelposition
Jan-Mathijs Schoffelen - 2011-07-07 11:01:50 +0200
Hi Diego, I think I solved it. I had to make a few changes to ft_componentanalysis and ft_rejectcomponent, among others. If I repeat the steps you take (i.e. ft_denoise_synthetic, ft_componentanalysis, ft_rejectcomponent, ft_megplanar, ft_combineplanar) all seems fine in terms of reconstructed sensor positions (which would mean no nans anymore). Unfortunately, you need to go back quite a few levels in your analysis pipeline to validate this. I would appreciate it if you coulr validate my changes with the 2 datasets which did not work for you. Could you let me know whether it works again? Thanks JM PS1: I am about to commit the changes, they will be available in about 20 minutes on /home/common/ PS2: You can always still work with the 'bad' data by replacing the data.grad with the original one, prior to calling ft_megplanar.
Jan-Mathijs Schoffelen - 2011-07-08 14:55:58 +0200
Created attachment 89 test data.grad
Jan-Mathijs Schoffelen - 2011-07-08 14:58:02 +0200
Created attachment 90 test data.grad
Jan-Mathijs Schoffelen - 2011-07-08 14:59:16 +0200
(In reply to comment #4) > Hi Diego, > > I think I solved it. I had to make a few changes to ft_componentanalysis and > ft_rejectcomponent, among others. If I repeat the steps you take (i.e. > ft_denoise_synthetic, ft_componentanalysis, ft_rejectcomponent, ft_megplanar, > ft_combineplanar) all seems fine in terms of reconstructed sensor positions > (which would mean no nans anymore). > Unfortunately, you need to go back quite a few levels in your analysis pipeline > to validate this. I would appreciate it if you coulr validate my changes with > the 2 datasets which did not work for you. Could you let me know whether it > works again? Thanks > > JM > > PS1: I am about to commit the changes, they will be available in about 20 > minutes on /home/common/ > PS2: You can always still work with the 'bad' data by replacing the data.grad > with the original one, prior to calling ft_megplanar. Hi Jan-Mathijs, I re-did my analysis from the very begining (ft_preprocessing-> ft_componentanalysis -> ft_rejectcomponent -> ft_rejectartifact -> ft_megplanar) and when the planar calculation I get this error: %planar calculation cfg = []; cfg.planarmethod = 'sincos'; cfg.channel = {'MEG'}; cfg.trials = 'all'; meg = ft_megplanar(cfg, meg); the input is raw data with 274 channels and 370 trials undoing the invcomp balancing Warning: the linear projection for the montage is not full-rank, the resulting data will have reduced dimensionality > In ft_apply_montage at 76 In fieldtrip/private/undobalancing at 29 In fieldtrip/private/channelposition at 31 In fieldtrip/private/megplanar_sincos at 26 In ft_megplanar at 171 undoing the comp balancing Warning: the linear projection for the montage is not full-rank, the resulting data will have reduced dimensionality > In ft_apply_montage at 76 In fieldtrip/private/undobalancing at 29 In fieldtrip/private/channelposition at 31 In fieldtrip/private/megplanar_sincos at 26 In ft_megplanar at 171 ??? Error using ==> ft_apply_montage at 133 not all output channels of the montage are unique Error in ==> undobalancing at 29 sens = ft_apply_montage(sens, sens.balance.(sens.balance.current), 'inverse', 'yes', 'keepunused', 'yes'); Error in ==> channelposition at 31 sens = undobalancing(sens); Error in ==> megplanar_sincos at 26 [pnt, ori, lab] = channelposition(grad); Error in ==> ft_megplanar at 171 montage = megplanar_sincos(cfg, data.grad); In ft_apply_montage at 133 seems to be a matrix dimension problem in the grad.tra structure but I can't figure out why. I attach you my grad matrix.