SignalConcatenation and StreamSwitch

[lbotrel]

Hello, i just subscribed to the forum.

I've been starting my PhD in Würzburg, Germany and successfuly translated a BCI2000 Cursor Task feedback paradigm to an OpenViBE scenario with all the proper stimulations.

First I used Olivier Rochelle BCI2000 translation box that works pretty well. Its output has 2 signal : 1 for EEG signals, and 1 corresponding to the variable used (TargetCode, Feedback, PositionY...).

I translated these signals by a box I created, that i called Signal Trigger, that allows to send specific stimulations when signal has a specific value.
Then i had problems to add start and end_of_session stimulations in my signal, so I created a begin- and end stimulation bounding box (adding stimulations when signal input recieves headers, and end chunk).

I noticed that in Signal Concatenation box (thanks for previous changes btw), the headers and end chunk were not removed between parts of the concatenated signal.

I would like to know if it's normal or not.

I also noticed something with the stream switch : is it normal that it can switch almost every kind of signal but not stimulations ?

I hope it could help you to improve these boxes.

Loic

[lbonnet]

Hi Loic !
Thanks for the feedback...
We already discussed this face to face at BCI workshop in Graz, but I guess the answer can interest more people around

First I used Olivier Rochelle BCI2000 translation box that works pretty well.

Good !
We will be glad to integrate this box as soon as it's totally stable. From what I recently heard it's on its way to release

I noticed that in Signal Concatenation box (thanks for previous changes btw), the headers and end chunk were not removed between parts of the concatenated signal.

I guess it's not normal, as one header should be enough (the box checks for coherency between all headers before starting the concatenation).
I will investigate this asap.

I also noticed something with the stream switch : is it normal that it can switch almost every kind of signal but not stimulations ?

The stream switch has been made for all kind of streamed matrices (ie. streamed matrices and its derivated; signal, feature vectr, etc.).
It's not a huge job to add stimulation stream as a possible input/output I guess
If you or someone else is willing to contribute a patch, I will gladly integrate it!

Cheers

Laurent-

[lbotrel]

Thanks for the answer lbonnet,
Once these problems are fixed I will make scenarios that are appropriate and which can be shared.

Best regards
Loic

[lbonnet]

I just patched the signal concatenation box.

The Header and End duplication problem should be solved.

Now the box has a third output (stims) that sends the stimulation OVTK_StimulationId_EndOfFile when the concatenation is finished.
With a Player Controller you can now easily script your concatenations.

I didn't investigate the Stream Switch so far, please let me know if you already did some modifications

Laurent

[lbotrel]

I could send you the boxes and scenario from BCI2000 recordings to OpenViBE translation,
but the bci2000 reading box is stored on OpenViBE-Cortex repository.
So that make no sense at all if I send this now.
If you get it from Nancy, then I will send you a nice christmas present

Happy Christmas !

[lbonnet]

Thanks Loic!

I manage to get the BCI2000 File Reader from INRIA Nancy. It's available temporary in a branch (file-io/branches/wip-bci2000).
It's an official candidate for integration for the next release (this month).

If it passes the tests it's ok for us.

Happy new year!

Laurent