Offline Analysis tool MatLab error [solved]

[pieterkubben]

Hello,

When I try to run the OfflineAnalysis tool in MatLab (both R2010A and R2011A on Windows 7) I get this error:

Code: Select all

??? Error using ==> OfflineAnalysis>reloadSettings at 1458
Some of the files necessary are missing or have been corrupted.  Please
reinstall.

Error in ==> OfflineAnalysis>OfflineAnalysis_OpeningFcn at 265
settings = reloadSettings(handles);

Error in ==> gui_mainfcn at 221
    feval(gui_State.gui_OpeningFcn, gui_hFigure, [], guidata(gui_hFigure),
    varargin{:});

Error in ==> OfflineAnalysis at 65
    gui_mainfcn(gui_State, varargin{:});

To avoid that is is related to any path settings I went to the corresponding directory first before opening the tool. Tried installing MCR afterwards, but it reports on a missing DLL (mclmcrrt78.dll). I'd preferably work from MatLab directly, therefore my question: is anyone familiar with this error? Looks to me that it is the the OfflineAnalysis script, and not in MatLab itself.

I am using BCI2000 v 3.0.4 on Windows at the moment.

Thanks,
Pieter

[pieterkubben]

Never mind... I won't delete the post in case somebody else has this problem, but after unsuccessful attempts on my computer in MatLab R2010A and R2011A I could try on another computer using MatLab R2012A and the same version of BCI2000 (v 3.0.4). No errors there.

So it either has to do with my computer, or these MatLab versions. In any case it does not seem to be related to BCI2000.

[pieterkubben]

I could just try on a third computer: MatLab R2010A and BCI2000 v 3.0.0
Comparable error... I get the impression that the OfflineAnalysis tool might not be compatible with older versions of MatLab. Do not know if anyone else had the same issue.

[mellinger]

Thanks for reporting these problems.

There have been a number of different issues with OfflineAnalysis in previous versions of BCI2000. A new binary release of BCI2000 is coming out soon, which will have these issues fixed. For now, I suggest that you use the current version from SVN.
Make sure to obtain the latest version of both the OfflineAnalysis and tools/mex directories from below http://www.bci2000.org/svn/trunk. If you don't have an SVN client installed, you may use a standard web browser to download individual files.

Regards,
Juergen

[pieterkubben]

Thanks! I first installed R2012A on my computer, the same one that was causing the errors in R2011A and R2010A. Also R2012A did have these problems. So it is not related to the MatLab version.

Then I installed the OfflineAnalysis tool from the trunk, which works like a charm. However, it still does not solve this warning about the unknown method when attempting to draw a topography plot.

Another issue which I supposed is not related (although I do not know the solution yet) is why the r^2 values are very low (around 0.05) when doing the sensorimotor rhythm tutorial + analysis. I could understand if it were only for the EEG, but is is also when I use the signal generator instead of EEG. Any idea on this? I use no filter when acquiring the data (just high pass 0.1) and a CAR in the OfflineAnalysis.

All the best,
Pieter

[mellinger]

when I use the signal generator

Are you referring to the BCI2000 SignalGenerator source module, or to a physical signal generator attached to an EEG amplifier? For the r^2 values to be significant, there must be amplitude modulation present in the signal, i.e. the amplitude must differ significantly between conditions. How are you modulating the amplitude when using the "signal generator"?

-Juergen

[pieterkubben]

when I use the signal generator

Are you referring to the BCI2000 SignalGenerator source module, or to a physical signal generator attached to an EEG amplifier?

I am referring to the first, the BCI2000 SignalGenerator source module. You have a point, did not think of this but it makes perfect sense. The Emotiv EPOC does not use an amplifier as far as I know (unless it is hidden in the software or the USB stick, but I would not think so). The source module may just be emulating the signal, without any sort of amplification. Which means that in both cases I will get low r^2 values. Do I understand this correctly?

So, then my next step should be to try with a conventional EEG first, meanwhile looking for a way to amplify the signal from the Emotiv EPOC.

As a sidestep: is there a way to amplify the signal from the source module (using some kind of software setting)?

Thanks,
Pieter

[mellinger]

An EEG amplifier is an electronic device connected to the EEG sensors, which amplifies sensor signals from their very small voltages (Microvolts: muV=1e-6V) range to larger voltages (about 1 Volt) suitable for input into an AD converter. Modern amplifiers have AD converters built in.

The Emotiv EPOC is an integrated device that contains sensors, amplifier, and AD converter.

Amplification has nothing to do with r^2 values. To have high r^2 values, the amplitude (not to be confused with amplification) of the signal must covariate with the condition under which it is recorded. E.g., large amplitude for "right hand" trials, low amplitude for "left hand" trials.

When you use the SignalGenerator source module, you may vary signal amplitude by moving the mouse, as described in the "Getting Started" document located in BCI2000's top level directory. To get high r^2 values from a SignalGenerator session, move the mouse to locations that correspond to instructions on screen. E.g., move the mouse to the right edge during "right hand" trials, and to the left edge during "left hand" trials.

HTH,
Juergen

[pieterkubben]

This definitely helps, thanks a lot. I thought that signal amplification and amplitude were related...

When you use the SignalGenerator source module, you may vary signal amplitude by moving the mouse, as described in the "Getting Started" document located in BCI2000's top level directory.

Read it, and after reading the rest of your explanation, will try this. So far I have only been working on the "initial mu session", but will continue playing around a little more. Should be fine now, I hope.

Thanks again!

Pieter