About the UCSD paper !!

Forum for software developers to discuss BCI2000 software development
Locked
MazouzBCI
Posts: 5
Joined: 19 Jul 2011, 15:36

About the UCSD paper !!

Post by MazouzBCI » 30 Jul 2011, 18:43

Hello World !
I'm Adam Mazouz; final year student in IEEE ( Digital Electronics ).
I've been workin on EEG and BCI ( especially based on DSP units ) for my
final year project; till I read the published paper of Non-contact
sensoring by UCSD where they used a PIC 24 to control the converted signal
from an ADS 7685. My questions are :

1. Is it possible to attache the design to any sensor ( either wet-Ag/AgCl or dry sensor or woteva kind of sensor it is ) instead of the non-contact capasitive sensor ?

2. Conserning the DsPIC24 program : does it only containt the
DATA-CONVERT-CLOCK-BLUETOOTH tags?! or are standards of feature extraction
and classification embeded ?

3. I wonder what software interface the UCSD use? ( is it the EEGLab ? ); if not , please mention it and the ability of getting it !!!!!!

4. Whare are the feature extraction and classification most used these
days ( especially by Neurosky and Emotiv )?

Really sorry for being a heavy guest !!

Thanks in advance
Best regards
Cheers!!
Adam

griffin.milsap
Posts: 58
Joined: 08 Jun 2009, 12:42

Re: About the UCSD paper !!

Post by griffin.milsap » 01 Aug 2011, 09:27

Hello Adam!!!!

Although I'm not entirely sure what your queries have to do with BCI2000 or specifically the software development thereof, I will do the best I can to answer a few of your questions.

1) By connecting this circuit to a wet, non-capacitive electrode you are effectively performing the amplification of the signal at the electrode site. This is referred to as an "active" electrode. By performing the amplification of a signal at the electrode source, a lot less noise is recorded - specifically when moving the wires which connect the electrode to the amplifier. Several manufacturers produce active electrode systems, though they can become quite expensive quickly. Personally, I have experience with g.tec's g.GAMMAsys - http://www.gtec.at/Products/Electrodes- ... s-Features. I can tell you that these work quite well, though do come at a cost. Besides being more expensive, you sacrifice some frequency response and performance. This is true of all active electrode systems.

2) I would seriously doubt that they're performing classification and feature extraction on a per-electrode basis at the site of the electrode. More likely than not, they're logging the signal using some software (maybe BCI2000 with a custom Data Acquisition Module?) and performing classification and feature extraction on a computer (the paper doesn't say that they performed this analysis online or offline).

3) No idea here, but since you're on the BCI2000 forums, you should know that you could use BCI2000 to log from a system like this! See - http://www.bci2000.org/wiki/index.php/P ... ion_Module for guidance on implementing a custom signal source module once you've written some device drivers for your own implementation!

4) I'm familiar with the Neurosky and Emotiv devices -- specifically, I wrote the data acquisition modules for BCI2000 for both of these devices. The Neurosky Mindset is a passive capacitive electrode system (compared to the active capacitive electrode system in the paper you described). Its one channel capacitive electrode is very noisy, poorly placed, and really only good for picking up eye movements and forehead muscle movements. The Emotiv EPOC at least has a better montage and better electrode technology, but is still somewhat noisy. As far as feature extraction and classification goes, the Neurosky uses simple filtering (band-pass, low-pass, high-pass, etc) and measures power at certain frequencies to determine "meditation" and "attention". These frequencies are proprietary and are not disclosed by the API, but it is also possible to read out the raw signal and perform your own feature extraction and classification in BCI2000. The Emotiv ships with a much more sophisticated learning algorithm based software which can be trained to implement a BCI. I assume it's really only doing a spectral analysis, but it's a bit harder to say with any certainty what it's doing.

That said, capacitive electrode technology has yet to show that it can seriously contend with traditional "wet electrode" technology. The signal is typically noisier and has a worse frequency response than the wet electrodes. For EEG the frequency response may not be as much of a problem.

Welcome to the BCI2000 forums,

-Griff

MazouzBCI
Posts: 5
Joined: 19 Jul 2011, 15:36

Re: About the UCSD paper !!

Post by MazouzBCI » 03 Aug 2011, 09:32

Hello Griff

That was the most fulfilling answer I had !!

And sorry; couldn't find a suitable place to put this on..

I really appriciate it !! thanks a lot mate

cheers!
Adam

Locked

Who is online

Users browsing this forum: No registered users and 3 guests