OpenBCI_GUI, even channels replaced with noise when file played back
Hi All,
I did not really know where to post this. So when it is not meant to be here please relocate.
The last day of the year I got time to do a simple experiment. I put electrodes on F3 F4 C3 C4 P3 P4 and O1 O2 I used white as SRB (Ref A1 + A2
I hoped to get some EEG data for playing around with. So I recorded 2 minutes Eyes Open, 2 minutes Eyes Closed and 2 minutes watching movie.
When I played back the data I got a different view as with the recording. See pic2
and 3
As you can see the even numbered channels have changed. You can also see that the 50Hz disturbance is gone for those channels.
So I opened the file and have a nice header with a "%". See below the first 6 rules.
%OpenBCI Raw EEG Data
%
%Sample Rate = 250.0 Hz
%First Column = SampleIndex
%Other Columns = EEG data in microvolts followed by Accel Data (in G) interleaved with Aux Data
24, -27009,76, -60278,41, -82223,98, -68034,77, -60148,34, -53729,95, -72783,27, -74142,27, -0,01, -0,01, 1,05
This looks fine, I guess. When I open the example file "openBCI_2013-12-24_relaxation.txt" the first rules look like this.
%OpenBCI Raw EEG Data % %Sample Rate = 250.0 Hz %First Column = SampleIndex %Other Columns = EEG data in microvolts
-29, 8081.03, -865.44, 19319.28, 31468.30, 4714.09, 36761.35, -587.38, 26779.51
-28, 7641.59, -845.37, 18335.31, 29751.33, 4410.33, 34759.82, -586.38, 25320.54
-27, 11011.34, -1218.13, 26441.22, 42836.22, 6349.84, 50050.63, -838.01, 36456.92
-26, 11253.32, -1239.29, 27008.26, 43812.66, 6513.92, 51190.20, -847.11, 37284.79
This file seems to playback fine.
Differences:
-The sample index is with a minus in the example file. Counting the other way around.
-The last 3 columns lack in the example file.
So where do I go wrong? Hope to hear soon of you.
Warm greetings and best wishes for 2015,
Paul
I did not really know where to post this. So when it is not meant to be here please relocate.
The last day of the year I got time to do a simple experiment. I put electrodes on F3 F4 C3 C4 P3 P4 and O1 O2 I used white as SRB (Ref A1 + A2
Trying to manage "Linked Ears" See pic1
I hoped to get some EEG data for playing around with. So I recorded 2 minutes Eyes Open, 2 minutes Eyes Closed and 2 minutes watching movie.
When I played back the data I got a different view as with the recording. See pic2
and 3
As you can see the even numbered channels have changed. You can also see that the 50Hz disturbance is gone for those channels.
So I opened the file and have a nice header with a "%". See below the first 6 rules.
%OpenBCI Raw EEG Data
%
%Sample Rate = 250.0 Hz
%First Column = SampleIndex
%Other Columns = EEG data in microvolts followed by Accel Data (in G) interleaved with Aux Data
24, -27009,76, -60278,41, -82223,98, -68034,77, -60148,34, -53729,95, -72783,27, -74142,27, -0,01, -0,01, 1,05
This looks fine, I guess. When I open the example file "openBCI_2013-12-24_relaxation.txt" the first rules look like this.
%OpenBCI Raw EEG Data % %Sample Rate = 250.0 Hz %First Column = SampleIndex %Other Columns = EEG data in microvolts
-29, 8081.03, -865.44, 19319.28, 31468.30, 4714.09, 36761.35, -587.38, 26779.51
-28, 7641.59, -845.37, 18335.31, 29751.33, 4410.33, 34759.82, -586.38, 25320.54
-27, 11011.34, -1218.13, 26441.22, 42836.22, 6349.84, 50050.63, -838.01, 36456.92
-26, 11253.32, -1239.29, 27008.26, 43812.66, 6513.92, 51190.20, -847.11, 37284.79
This file seems to playback fine.
Differences:
-The sample index is with a minus in the example file. Counting the other way around.
-The last 3 columns lack in the example file.
So where do I go wrong? Hope to hear soon of you.
Warm greetings and best wishes for 2015,
Paul
Comments
Regarding the 1st column of data, this is just an incrementing packet counter. So any values (even negative looking), as long as they keep adding 1 to the next row, are fine. The 2013 relaxation.txt recording was made some time ago, before the accelerometer values were being recorded. Their absence should just be ignored anyway.
What you point out with the noise on the even channels on playback looks like a bug to me. I believe that Conor will be taking a look at this. Can you email your recorded file to [email protected] , attention Conor?
The OPENBCI_GUI program was a collaboration amongst various authors: Conor, Chip, etc. I believe Conor has done the more recent changes. I'm listing both their usernames here so they should get an email notification: @conor_obci and @chipaudette .
Is there any possibility that before doing your recording, you had changed some of the default settings in the GUI, that could have caused test signals to be recorded on those even channels? You might also try a recording with say just channels 1 and 2. Starting up fresh with all settings at default. In the example you show, where was your Bias located? It's not that critical, but some EEG experts suggest a midline location (e.g. AFz) for the bias has a more even bilateral noise cancelling effect.
William
Thanks for the response.
As far as I am aware of it I did not change any settings. Mail Send. BIAS was on Cz.
Warm greetings,
Paul
http://openbci.com/forum/index.php?p=/discussion/69/negative-sampleindex-repeating-29-to-0
My impression from reading Chip's comments there is that the negative sampleindex issue may have been fixed at one time. Because the firmware now does fill in that field correctly. I've seen some other output from the GUI (a month or so ago), where the sampleindex was cycling correctly through positive values.
William
I am not well enough informed to know IF the negative values of the sampleindex field could be a problem. Probably not, since that file is playing back fine. It's the strange behaviour of the even channels that makes me unsure a bit.
BTW I repeated the process in Brainbay and that seems to work fine. But I have not found a way yet to open those files in a readable format.
Warm greetings,
Paul
The EEG block in Brainbay will let you save the file in "Integer Values" (Brainbay internal) mode, this allows for playback. Some of the demo designs use this.
More interesting may be the EDF Writer block under Targets. This records in the EDF standard. Which can be browsed with the EDFBrowser free app.
http://www.teuniz.net/edfbrowser/
Cheers, William
PS the negative sampleindex is just a minor glitch and is not related to the even channels behavior.
Is there any way that you can share the full data file with me? Dropbox, google drive, github, somewhere?
I'd love to take a look at the full file to see what your problem might be...
Chip
I mailed a file to @conor_obci. Can you share or should I mail you a copy. I am in a bit of a rush right now. But tomorrow I would have time to do this. Can I use the mail adres from EEG hacker?
Warm Greetings,
Paul
I'd forgotten about this. Sorry!
I took a first look at your data and I confirmed that something is definitely weird about your even numbered channels. To see the data, you can view an HTML version of my IPython Notebook:
http://nbviewer.ipython.org/github/chipaudette/EEGHacker/blob/master/Python/2015-01-17%20Missing%20Even%20Channels/exploreData.ipynb
The time-domain plots clearly show that the even channels have nothing going on. And the spectrograms that follow show just noise for the even channels.
More to follow...
Chip
Cheers,
William
Warm greetings,
Paul
PS Am I the only one who observes this??