bci2000.org BBS

Hello,

I am doing a BCI project using BCI2000. The task is i need to find out the total beta1/Theta ratio from the 14 channel output that I am getting from EMOTIV tool kit. As I am a new one in BCI2000 Its make more complications in the programming. I Succesfully compiled the EMOTIV ADC source and put it in the signal source column. I think that the next thing is that to create a signal processing module and define the functions and programming needed.I am planning to do ARfiltering for 14channels and finding out the Beta1 and theta power spectrum for all the channels seperatly and using linear classifier to sum up all the channels beta1 and theta power into 2 separate channels. After that I am planning to divide those channels using expression filter to find the beta1/theta ratio of the overal channels. Please inform me wheather i am correct or not? And I don't have any Idea that hoe to use ARfilter for seperate 14 channels and how to define the particular spectrum(in my case for each channel beta1 and theta).Can I define all these using C++ language, with the source code available? I need your kind advice for completing my project.


thanking you in advance

Aswin Venu

I only have my macbook with me so I can't easily check the parameters, but I believe you would want to set the frequency range and window width to be sufficient to capture both theta and beta1. Then, using linear classifier, you'd multiply all your features that fall in the theta range by 1/(14*N) (where N is the number of features in the theta range) for output1 and similarly for beta1 for output2. Then, as you said, use the expression filter to divide output2/output1.

Also note that the pre-built application modules expect the feature to have an offset of 0 and a variance of 1. Beta1/Theta will have a non-zero offset with a variance much smaller than 1. If you plan to use a pre-built application module then you may also want to use the Normalizer filter in the SpectralSignalProcessing module.

There already exists a module that does exactly what you describe: SpectralSignalProcessing. You can achieve what you describe by correctly parameterizing the already available SpectralSignalProcessing module.

Thank you very much for your answer. From your answer I can understand that, Using already available signal processing module-Spectral signal processing module i can do the all! only by giving the correct parameters values. Once again thank you very much for your reply.

Aswin Venu