Back to the main page.
Bug 2140 - discrepancy between trials/time when filtering in-memory trials with mirror padding
| Status | CLOSED FIXED |
| Reported | 2013-04-26 11:57:00 +0200 |
| Modified | 2014-03-12 12:21:33 +0100 |
| Product: | FieldTrip |
| Component: | core |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Linux |
| Importance: | P3 normal |
| Assigned to: | Jörn M. Horschig |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: |
Robin - 2013-04-26 11:57:07 +0200
Is it possible filter in-memory extracted trials data with mirroring at the edges? When I try: % filter cfg = [];cfg.lpfilter = 'yes'; cfg.lpfreq = '40'; cfg.lpfilttype ='but'; cfg.padtype = 'mirror'; % trial length is 1.1s so mirror-pad to1.5s (200ms each side) cfg.padding = 1.5; flt_data = ft_preprocessing(cfg, block_data); I get: Reference to non-existent field 'Fs'. Error in ft_preprocessing (line 288) padding = round(cfg.padding * data.Fs); Adding block_data.Fs = block_data.fsample; seems to work. But in the resulting data structure the time vectors no longer match the trial data, so obviously something is going wrong: >> size(block_data.time{1}) ans = 1 560 >> size(block_data.trial{1}) ans = 234 560 >> size(flt_data.trial{1}) ans = 234 560 >> size(flt_data.time{1}) ans = 1 357 This is using SVN r8075. block_data is the result of ft_appenddata of several individual runs.
Jörn M. Horschig - 2013-04-26 12:26:37 +0200
Created attachment 468 corrected for in-memory data - is this correct for data on disk as well? Hi Robin, in the meanwhile I found that the error is because the number of padded samples also get deducted from the time-axis, while time-axis did not get padded. So that was an easy fix, see attachment. I just wonder whether the same fix that I applied for data in memory has to be applied for data on disk. The attached version applies the same fix for data on disk, but I am not sure whether this is the correct thing to do. I will still have to do some sanity checks for this. However, your bug should be fixed with this version.