Using GRASS amps in initial session

XiangL · Post by **XiangL** » 07 Dec 2009, 11:58

Hi All,

I am trying to running a initial session to get Mu Rhythm Parameters. And I am wondering if GRASS amplifiers could be used in this session. If yes, how could I configure the it through BCI2000, because I don't see any ".bat" file named as "StimulusPresentation_Grass.bat" in the "batch" folder.

Thanks a lot!

Regards,

Xiang

mellinger · Post by **mellinger** » 08 Dec 2009, 08:47

Hi Xiang,

currently, there is no support for Grass amplifier/digitizer systems in BCI2000. AFAIK, Grass does not offer real-time access to acquired EEG data, so it is difficult to add support for their amplifiers.

A list of contributed BCI2000 source modules which add support for a number of amplifiers/digitizers can be found in the wiki at
http://www.bci2000.org/wiki/index.php/C ... tions:ADCs

HTH,
Juergen

XiangL · Post by **XiangL** » 15 Dec 2009, 11:43

Dear Juergen,

Thank you for your nice guide, we have been making progress in recent 3 days.

and now we are trying to generate a subject-specific parameter file. We have gone through calibration and acquired the data files for the "the quick brown fox". We then did offline analysis and generated plots. We selected 5 strong points from the plot and put them in the "Waveform Channels" and "Topo Times". Next we changed the filter in the config file to a 4 by 5 matrix and entered the values (of course we didn't include the ms at first and kept getting nasty red text). Then we performed the calibration again (we actually did it three more times).

What we are unclear about is when we do the offline analysis again and it gives us different times, and possibly different channels, what do we include in the filter? Do we keep the values from the previous analysis and add the new times and channels or do we only enter the new values?

Also, when we perform the offline analysis should we include all the files from each calibration or only the most recent files?

Thanks a lot!

Xiang

gschalk · Post by **gschalk** » 24 Dec 2009, 10:21

Xiang,

You do not write whether you can successfully spell using your setup. Let us know.

What in general you need to do is to analyze different experimental runs to see which features are the most stable. The easiest procedure is probably to simply use a larger dataset, and then pick the best features from that.

Please also note that the P300 GUI program, and the P300 Classifier that comes with the upcoming BCI2000 V3.0, does the feature selection automatically. Generally, the use of these programs also leads to better results (i.e., accurate spelling at fewer stimulus repetitions). Analyzing the data by hand is still informative to learn about the P300 response, though.

Gerv

XiangL · Post by **XiangL** » 26 Dec 2009, 22:55

Xiang,

Hi Gerv,

Thank you for the reply. Now we could successfully spell using our configuration. But we still have two questions.

1. When we configure the BCI2000 in p300 experiment, there is an item for us to input the DAQ range. If we input 5v -5v, for example, does this mean we tell the BCI2000 that the DAQ range is +/- 5 volts or this BCI2000 will set the DAQ range to +/- 5 volts?

2. In the BCI2000, could we combine two channels of input signal somehow and get the combined data normally processed as one channel?

Thanks a lot!

Xiang

What in general you need to do is to analyze different experimental runs to see which features are the most stable. The easiest procedure is probably to simply use a larger dataset, and then pick the best features from that.

Please also note that the P300 GUI program, and the P300 Classifier that comes with the upcoming BCI2000 V3.0, does the feature selection automatically. Generally, the use of these programs also leads to better results (i.e., accurate spelling at fewer stimulus repetitions). Analyzing the data by hand is still informative to learn about the P300 response, though.

gschalk · Post by **gschalk** » 27 Dec 2009, 07:27

Xiang,

I am glad you are successful in your experiments.

I need to know which Source module you are using to be able to answer your first question.

I do not understand the second question. Do you mean raw EEG channels or classifier outputs?

Gerv

XiangL · Post by **XiangL** » 31 Dec 2009, 09:33

Hi Gerv,

For the first question, we are now using P3speller_DASSource, The DAQ is USB-2537 from Measurement Computing.

For the second question, we want to combine the raw data received from the EEG channels.

Thanks,

Xiang

mellinger · Post by **mellinger** » 04 Jan 2010, 07:47

Xiang,

ad 1)
the DAS source module will configure your board according to the selected input voltage range. Optimal settings depend on what your board supports, and on the amplifier's output voltage range.

ad 2)
EEG channels may be linearly combined using the spatial filter, described at http://www.bci2000.org/wiki/index.php/U ... tialFilter. However, you cannot use this combination in the online analysis tool, so it might be better to keep the channels separate, and only combine them in the linear classifier. For the P300 signal processing, it does not matter whether channels are combined in the spatial filter, or in the linear classifier -- both will result in a linear combination of signals.

HTH,
Juergen

XiangL · Post by **XiangL** » 04 Jan 2010, 20:03

Hi Juergen,

Thanks a lot! The problem now is that when we input the DAQ range as "-5 5", at the location: " Source -> ADRange", and click the "Set Configure" button, it gives out the error information as:" TDAS_ADC:: Preflight: unhandled exception EInvalidOp, Terminating SignalSource module." But if we leave the range as "-10 10", which we usually use, it works fine.

Best,

Xiang

gschalk · Post by **gschalk** » 05 Jan 2010, 07:04

Hi,

If you set the input range to 5V, BCI2000 should set that range in the amplifier. I would guess that the source module crashes if you use 10V, because the board probably does not support +-10V input range. In general, which input range you use depends on your amplifier and resolution of your DAS board. If you only have 12-16 bit resolution, you probably want to set the input range such that, before amplification (which depends on the amplification of your amplifiers), you have about 100-200muV input range.

Gerv

XiangL · Post by **XiangL** » 08 Jan 2010, 17:18

Hi, Gerv

We checked the DAQ again, it does support both +/-5 and +/-10 input. For BCI2000 configuration, +/- 10 is the default range, it works fine with our DAQ, but if we change it to +/-5, it would give the error I mentioned above...... The want to reduce the input range because the signal amplitude is not very strong.

Best,

Xiang

mellinger · Post by **mellinger** » 27 Jan 2010, 08:31

Xiang,

without having the hardware available, we unfortunately cannot reproduce your problem with the +-5V input range.

You could assist us in trying to find out what is going on by running the source module in the Borland IDE debugger. It will stop at the location in the source code where the EInvalidOp exception is thrown. You can then report the file name and line number where it happens.

Juergen

XiangL · Post by **XiangL** » 27 Jan 2010, 09:54

Hi Juergen,

By changing the range setting in source code (TDAS_ADC.cpp in DASSource.cbproj), now we are able to change the range to what ever the MMC USB-2537 supports. But there is still some problem with MMC USB-2416-4AO DAQ. We will try to fix this.

Regards,

Xiang

bci2000.org BBS

Using GRASS amps in initial session

Using GRASS amps in initial session

P300 ...

DAQ

DAQ

Who is online