How to Test P300 Magic Card using Emotiv EPOC headset

[davidchung]

I would like to test p300-magic-card. I am assuming that the procedure should be similar to p300-speller-xDAWN, but I may be wrong. After running the p300-magic-card, I have some questions.

Step 1. I ran "p300-magic-card-1-acquisition". I have focused on a card highlighted with "green". Unlike "P300-speller-1-acquisition.xml (in p300-speller-xDAWN), the acquisition process was much longer than xDAWN's acquisition phase.

Step 2. Next, I ran "p300-magic-card-2-train-classifier", however, I only see the 4 x 4 magic card matrix screen, but no random flashing in order to train the classifier. What is happening here?

Step 3. I am guessing that the final step is to run "p300-magic-card-3-online". Is that it?

]Please advice me about the procedure, if I am doing wrong here.

Thanks,
David

[bpayan]

Dear davidchung

You're right, the p300-magic-card is similar as the p300-speller-xDAWN, but you haven't the step for learn the spatial filter.

The first step is the acquisition with the scenario p300-magic-card-1-acquisition. The target is the picture with a green border.

The second step is the training, the acquisition file is loaded with the “Generic stream reader” box and the “Classifier trainer” box make a file configuration with information about the classification model.

For your problem, without log messages, I think the Generic stream reader wasn't configure to load a good file. Can you post your log file, it will be a good start to find the reason of your problem.

And the last step is the online, it's similar as the first step, but after a trial, the result is the picture with a red border.

I hope this helps.

bpayan

[davidchung]

The first step is the acquisition with the scenario p300-magic-card-1-acquisition. The target is the picture with a green border.

ok.

The second step is the training, the acquisition file is loaded with the “Generic stream reader” box and the “Classifier trainer” box make a file configuration with information about the classification model.

For your problem, without log messages, I think the Generic stream reader wasn't configure to load a good file. Can you post your log file, it will be a good start to find the reason of your problem.

I still have the same problem where Step 2 does not do anything. The acquisition file is loaded with the Generic stream reader box (e.g. current-[2011.01.10-10.23.43].ov). FYI, the logs are attached (openvibe-acquisition-server and openvibe-designer).
Please let me know what I did wrong. Thanks!

A quick question. Why doesn't the "magic card" scenario require training of spatial filter? Isn't the magic card scenario using the same P300 BCI technique as the P300-speller-XDAWN?

Again thanks,
David

[bpayan]

Dear davidchung

Thank you for your logs. I looked them and it seems that no box had a problem, so it's possible the problem come to the data of your file. I think you had tried the step 2 with the default file “bci-p300-magic-card.ov“ and all was right. Could you send me your file “current-[2011.01.10-10.23.43].ov” and your scenarios “p300-magic-card-1-acquisition” and “p300-magic-card-2-train-classifier”, I'll try to find the reason of your problem. What is your version of OpenViBE?

A quick question. Why doesn't the "magic card" scenario require training of spatial filter? Isn't the magic card scenario using the same P300 BCI technique as the P300-speller-XDAWN?

This scenario was created before the development of the xDAWN box, so these scenarios was made without it. But you are right the magic card use the same BCI technique as the P300-speller-xDAWN. That's why it's possible to modify the current scenario and adding the xDAWN function.

Best regards,

bpayan

[davidchung]

Thank you for your logs. I looked them and it seems that no box had a problem, so it's possible the problem come to the data of your file. I think you had tried the step 2 with the default file “bci-p300-magic-card.ov“ and all was right. Could you send me your file “current-[2011.01.10-10.23.43].ov” and your scenarios “p300-magic-card-1-acquisition” and “p300-magic-card-2-train-classifier”, I'll try to find the reason of your problem. What is your version of OpenViBE?

The attached are logs. I am currently using OpenViBE 0.80 installed in Windows XP machine.
The forum allows max. of 250KB per file. My file, current-[2011.01.10-10.23.43].ov, is too big even after compression. Sorry...


I would like to test the Magic Card scenario using P300-xDAWN algorithm with "different matrix size".

In "p300-magic-card-1-acquisition", I have noticed that the value of "Number of rows" and "Number of columns" are both 6, respectively. Why aren't they "3" and "4", respectively?


Thank you!
David

[bpayan]

Dear davidchung,

I used your scenarios and I hadn't got problem to run them. In the second scenarios the cards are flashed, but I have made few modifications in the first scenarios for this.

This scenario has two running modes and it is composed to three part.
The first is a read file, to replay a recording session.
The second is an acquisition branch, at right.
The third is a visualisation and filter part, at the down of the scenarios. (identity, Channel selector, P300 Magic Card Visualisation, …)
The identity box at the down, is used to switch between the acquiring mode and the replay mode.
If you used this scenarios without modification, you are in the replay mode, that's why you see that cards are flashed but this is just a replay and the file you obtained after that, has just signals information recorded, because the real application wasn't launched.
If you connect the two identity's boxes together, you use the acquiring mode and to launch the application you need to press a key (example “a”), in the window “keyboard stimulator”.

In "p300-magic-card-1-acquisition", I have noticed that the value of "Number of rows" and "Number of columns" are both 6, respectively. Why aren't they "3" and "4", respectively?

If you compare this scenario with the first scenario of P300-speller, you can see that only one box are different, “P300 Speller Visualisation” and “P300 Magic Card Visualisation”. The Magic Card scenario is based on the P300 speller, only the visualisation of the application is different. Now for answer your question, The P300 speller flash rows and columns but the Magic cards flash each cards one by one, so for the same dimension matrix, you haven't the same number of flash. The “P300 Magic cards visualisation” doesn't used the stimulation, given by the “P300 speller stimulation”, as the “P300 speller visualisation”. It use the rows stimulation for the first cards and the columns stimulation for the last cards. In your case you have 12 cards, so the rows stimulation are used for the 6 first cards and the columns stimulation for the 6 last cards.

I would like to test the Magic Card scenario using P300-xDAWN algorithm with "different matrix size".

Warning, in the "P300 speller stimulation" values of row and column must be same.

For changed the visual matrix dimension, you can made this changed in the dist/share/openvibe-plugins/simple-visualisation/p300-magic-card.ui
but the box isn't create for exceeded 12 cards, you can increase this value if you made a few modifications in the source code.

I hope this helps, and we are sorry for the inconvenient, these are old scenarios and we haven't made update as P300 speller xDAWN scenarios, that's why they are not easy to understand them.

Best regards,

bpayan