bci2000.org BBS

Hi,

i am an undergraduate, and i am developing a BCI based communication tool for disables. Since my project is self funded, i had to settle with a Neurosky mindset (couldn't go for a gtec or Emotiv head gear due to a financial issues).

Any way i was able to create a BCI that is controlled with Eye Blinks, and users are capable of sending a Tweet, Email and perform a Google search with it.

I need to evaluate my application, so i hope all the researchers from this forum would take a look at my application and comment on it. I am yet to create a video of how it functions, and i'll post it to this forum in the next few weeks.

I went through this http://www.youtube.com/watch?v=205dHV55XWQ video . this project was done by gerwin schalk, adam wilson and justin william. It demonstrates a person tweeting a sentence "SPELLING WITH MY BRAIN". I would like to test my application by tweeting a similar sentence.

I have some questions on how Gerwin and team tested this application,

1.) Has the participant gone through any training prior to this?
2.) How many attempts should we give a participant to do a particular task?
3.) Do i need to educate the participants on what the system is, and how it functions ? (will that have an impact on the result)

thank you, and looking forward to hearing from you

_JK

Hello,
This is Adam Wilson, from the video you posted. I did that experiment at the University of Wisconsin using BCI2000, and can answer your questions.

1. I was the subject in that video, so I had done some prior training. However, the nice thing about P300-based BCI applications is that they require very little training (unlike the mu rhythm cursor task).

2. Typically, we will configure BCI2000 to flash each row and column 15 times to select a letter the first few times the subject tries it. BCI2000 has a P300 analysis program as well that will tell you approximately how many flashes are required for 100% accuracy. In some subjects, this can be lower than 5 flashes, which is pretty fast. In that video, I used 10 flashes to make sure there was 100% accuracy for the video.

3. It helps if they understand a little about what they are doing. I will explain the basics of the P300 response, like: "Every time the letter you want flashes and you count to yourself, there is a change in your brain activity. When there have been enough flashes, the computer figures out which letter you were paying attention to by seeing which letter elicits the biggest change in brain activity." Or something like that. This can help them pay attention and understand the importance of counting, etc.

Let me know if you have any other questions! We are looking forward to seeing your program.
Adam

Hello Adam Wilson and the rest,

The application that i developed isn't a P300 based one, actually i worked on a P300 system but i couldn't find a suitable hardware as my project has no funding in any sort. So considering the budget factor i had to settle with a Neurosky device. I have used eye blinks to establish communication.

I have a few more questions regarding your project.

1.) How long did you take to type the sentence "GO BADGERS" and "SPELLING WITH MY NAME"?
2.) How did you calculate the error rate? is it ( the number of incorrect characters / total number of characters ) * 100
3.) By the look of it it seems you have used a G.tec EEG acquisition. Is it correct?
4.) Have you tested your application with a Locked-in patient?

I did an end user survey to check the average cost a person wish to spend on such a BCI application/device and the results was < 200$ - 400$. Since i am making use of a Neurosky mindset in this project, the overall cost is less while comparing with a commercial EEG/BCI project.

This is the projects Twitter profile http://twitter.com/#!/Dexbrain
Looking forward o hearing from you.

regards

jk