OpenBCI: connect your brain to your computer!

The subject area and the goals are interesting, but whoever thought up the "artistic" positioning of connectors in a semicircle deserves a severe whipping.  Clearly someone thinks that it's an arts project, and engineering commonsense be damned.

Morgaine.

It's an interesting TI chip they're using. I just skim-read the kickstarter page, and couldn't tell the reason for the board shape, but I think maybe they expect to mount it inside a cap, rather than a normal box, so maybe that's the only reason they've got is arranged like that. I too think a normal shape, separate box would have been a better idea, since it would be more discreet than a cap with electronics sticking out, and you might want it in a larger screened box. But maybe I've misunderstood some stuff about the project. It's a lot of money for home use (I know a medical device would cost many times more!) but maybe it can be done for lower cost with an implementation that used a BBB/RPI for the data collection and processing (expecially since there must be many math applications in Linux).

This is what olimex has already made: https://www.olimex.com/Products/EEG/

To me it also looks too expensive for what it is. I'm missing a list of proven track records to support such high cost. I also skim-read, so could've passed some major highlights

shabaz wrote:

 

maybe it can be done for lower cost with an implementation that used a BBB/RPI for the data collection and processing (expecially since there must be many math applications in Linux).

That does seem a reasonable suggestion based on just a quick glance at the ADS1299's specs.  OK, so it's low noise and provides a lot of EEG-specific functionality at the probe interface, but ultimately it's a glorified 8-channel 24-bit ADC sampling in the audio range, and I suspect that vastly cheaper audio-oriented devices could be used for the task since 24-bit/192KHz is the commercial sweet spot in digital audio.

And as you say, there's no shortage of maths processing libraries available in Linux, for every language under the sun.

Morgaine.

shabaz wrote:

 

maybe it can be done for lower cost with an implementation that used a BBB/RPI for the data collection and processing (expecially since there must be many math applications in Linux).

That does seem a reasonable suggestion based on just a quick glance at the ADS1299's specs.  OK, so it's low noise and provides a lot of EEG-specific functionality at the probe interface, but ultimately it's a glorified 8-channel 24-bit ADC sampling in the audio range, and I suspect that vastly cheaper audio-oriented devices could be used for the task since 24-bit/192KHz is the commercial sweet spot in digital audio.

And as you say, there's no shortage of maths processing libraries available in Linux, for every language under the sun.

Morgaine.

OK, so it's low noise and provides a lot of EEG-specific functionality at the probe interface,

Well, those are exactly the problems that arise when processing biomedical data. So it's very sensible to use such an interface. Most audio solutions expect a low-impedance driver, which the human skin obviously isn't. Add to that a good CMRR and low leakage...

vastly cheaper audio-oriented devices could be used for the task since 24-bit/192KHz is the commercial sweet spot in digital audio

...which are mostly 2-channel. So you'd need 4 of those devices plus the preamps to get the good CMRR and low leakage.

I think it would be more worthwhile to make something with a good data interface (CAN? It's balanced and high-speed) and put the very noisy BBB and Rpi somewhere where it doesn't interfere with the measurement. Keep all analog stuff close to the source, and then traverse the distance with a digital signal. A small micro with a local power supply could do the acquisition & data transfer near to the measurement pod.

Oh no - I'm going to have to be a PITA again.

This is a nice project but it's abosulutely NOT SAFE !!!

There is a reason that medical electronics has to pass much more stringent specs for safety than a PC or lab power supply. That nice tight electrical coupling to your head is good for picking up little signals from your brain and equally good at killing you if something goes wrong. I may have missed it but I couldn't find any references to isolation on the Kickstarter pages.

Picking up on Victor's suggestion - if this thing was battery powered and used a wireless link to the BB or PI the little signals would be happy and the maker's brain would not risk being fried.

MK

Very well pointed out, Michael.

This is an area where professional responsibility must play a very important part, and that includes not only designing for complete safety but also to be totally idiot-proof.  OSHW is not immune from those considerations.  Caution is advised to the point of not doing it at all if it would entail taking safety shortcuts.

Does anyone know whether non-contact sensing could provide any worthwhile bio-feedback signals?  Extremely low level radiated electrical fields and capacitive and infra-red sensing are unlikely to provide the same bandwidth of information as contact EEG, but there would probably be some usable information there, and the risk far lower.

Morgaine.

Hi Morgaine,

There are some methods, but not as clear / deterministic as EMG or EEG. The amount of bloodflow can be measured both by laser doppler shift, and by measuring thermal step response (reaction to heating of skin with IR lamp, measure the amount of heat transported away with FLIR camera, see this: http://www.neurosipe.nl/project.php?id=26)

Thanks Victor for the "Project 10730: Beyond Pain" link.  I hadn't come across EMG before -- Electromyography.  There are a couple of low-cost EMG devices listed at the bottom of the Wikipedia article.

It seems that EEG can be done with non-contact sensors too --- here's a paper on a capacitively coupled probe, "A Low-Noise, Non-Contact EEG/ECG Sensor".

Morgaine.

EMG is also very weak, but a lot stronger than EEG. Most BCI say they use EEG, but most do actually use EMG, where your frowning( hard thinking to activate EEG) activates so many muscles that any EEG is swamped, but it the device does react to you! Take a look at the project of Cosmin Iorga in the PSOC challenge, he actually used his eye muscles to command his TV.... (EOG)