Back to the main page.
Bug 941 - ft_channelrepair doesn't reconstruct channel well
| Status | CLOSED FIXED |
| Reported | 2011-09-07 12:49:00 +0200 |
| Modified | 2011-09-14 14:33:50 +0200 |
| Product: | FieldTrip |
| Component: | preproc |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Windows |
| Importance: | P1 normal |
| Assigned to: | Jörn M. Horschig |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: |
Anne van Hoogmoed - 2011-09-07 12:49:11 +0200
I've tried to use ft_channelrepair to reconstruct a channel (channel 25). However, if I plot the results, the values of the repaired channel do not lie within the range of the values of the neighbours used to reconstruct the channel. Shouldn't that be the case? I've used the following script: data_eeg_clean.elec = elec_new; cfg = []; cfg.badchannel = {'25'}; cfg.neighbours = neighbours; data_eeg_repaired = ft_channelrepair(cfg,data_eeg_clean); cfg = []; cfg.channel = {'19','20','24','25','26'}; ft_databrowser(cfg, data_eeg_clean);
Anne van Hoogmoed - 2011-09-07 12:52:22 +0200
Created attachment 130 data
Robert Oostenveld - 2011-09-07 12:55:21 +0200
Thanks for the data. Someone will look at it soon. A test script should be made that demonstrates the problem. I have already copied the data to /home/common/.../data/test/bug941.mat
Jörn M. Horschig - 2011-09-12 12:39:37 +0200
Hey Anne, The fixed function should be on /home/common/ within a few minutes. The problem was caused by the default behaviour of the Matlab function 'setdiff'. This function sorts it's input alphabetically, leading to a different order for channels in the function than they are ordered in the data. Since your channels are called something like '1', '2', (...) '10', '11', (...), setdiff set the order to '1', '10', '11', (...), '2', '20', (...). Thus, the sensor indices did not match anymore and so the sensors taken for averaging were completely wrong. I fixed this by sorting the indices that setdiff returns numerically - so that the output is consistent again with the labeling in the data. @Robert: This behaviour might have caused wrong interpolation for others that use channelrepair, should we tell people(==mailing list) that there was a bug in this function?
Robert Oostenveld - 2011-09-12 13:12:15 +0200
(In reply to comment #3) > @Robert: This behaviour might have caused wrong interpolation for others that > use channelrepair, should we tell people(==mailing list) that there was a bug > in this function? Yes, good idea.