Back to the main page.
Bug 1856 - make sure that elec/grad channel units and that hdr.unit are specified for MEG systems, biosemi, brainvision and egi
| Status | CLOSED FIXED |
| Reported | 2012-11-27 16:42:00 +0100 |
| Modified | 2015-07-15 13:31:11 +0200 |
| Product: | FieldTrip |
| Component: | core |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Mac OS |
| Importance: | P3 normal |
| Assigned to: | Robert Oostenveld |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: |
Robert Oostenveld - 2012-11-27 16:42:39 +0100
these are the most common systems todo: - write test script and ensure that the units are not unknown
Robert Oostenveld - 2012-11-28 09:49:43 +0100
I have created a first version of a test script. I don't know whether it runs, because all test data is on another server. EGI is still missing from the list of file formats, I have to check whether the server contains data for that. mbp> svn commit test_bug1856.m Adding test_bug1856.m Transmitting file data . Committed revision 7008.
Jan-Mathijs Schoffelen - 2013-08-08 09:03:38 +0200
Test script fails for quite a few cases: -ctf64: low level problem reading the header -BTI: ft_chanunit returns unknown for gradiometers. According to the low-level header information this is most likely 'T', can be easily fixed -itab: in general the chantype does not conform FT-convention, e.g. magnetometers are 'mag' and references are 'ref mag' -itab153: the percentage of 'unknown' chanunits is 20.3%, (due to the high number of auxiliairy channels with unknown unit), the test script throws an error at 20% -itab28: the percentage of 'unknown' chanunits is 37.5% -yokogawa160 fails for reasons I have not yet investigated. Suggestion to fix part of the issues: update ft_chanunit (for bti) and ft_chantype (for itab). Shall I do that?
Vladimir Litvak - 2013-08-09 09:59:13 +0200
(In reply to comment #2) I can look at Yokogawa as I'm now working on that anyway. For the rest I'd be happy if it's fixed.
Robert Oostenveld - 2013-10-30 11:20:17 +0100
mac001> svn commit Sending fileio/ft_read_sens.m Sending test/test_bug1856.m Sending test/test_bug963.m Transmitting file data ... Committed revision 8675.
Robert Oostenveld - 2014-02-24 10:56:40 +0100
I closed several bugs at once that all have been resolved for some time. If you disagree, please reopen.
Eelke Spaak - 2015-02-11 14:02:06 +0100
For one dataset, the test script started failing: /home/common/matlab/fieldtrip/data/test/original/eeg/bdf/Newtest17-256.bdf
Vladimir Litvak - 2015-02-11 14:45:36 +0100
(In reply to Eelke Spaak from comment #6) OK. I'll check. I looked at those examples and it all worked fine but maybe I missed the 256-channel one. In any case the proper fix should probably be in ft_chantype and ft_chanunits. My PC is being changed today so it might take me till tomorrow to fix it. Vladimir
Vladimir Litvak - 2015-02-11 16:18:49 +0100
(In reply to Vladimir Litvak from comment #7) I made a fix and it should be OK now. This was just a dataset with 17 channels that couldn't be properly recognised. Not sure if we should really support it but as long as only Biosemi and ext1020 labels are used on Biosemi systems, it should be OK. I suggest you also add to that test script a Biosemi example with 1020 labels. Here is one: https://www.dropbox.com/s/8yp94805dtdqi4e/050327BH_overCZnoAlpha.bdf?dl=0 Vladimir
Robert Oostenveld - 2015-02-12 18:06:47 +0100
(In reply to Vladimir Litvak from comment #8) I added the bdf file to the test script, it works fine. mac011> svn commit test/test_bug1856.m Sending test/test_bug1856.m Transmitting file data . Committed revision 10230. EGI is still missing from the test script, but those formats are a mess ny themselves. I don't think we need them in this script, since the script demonstrates the core functionality.