SPEARS Algorithm and the iBrain

[Stefanj]

Hi All

Within the last few days a number of articles popped up in the media about a certain Dr. Philip Low that created the SPEARS algorithm. It is claimed that this algorithm allows a single channel of EEG data to be rapidly quantified, analyzed and classified. The article goes on to mention that together with the iBrain (single channel EEG amp device), it is hoped to give Stephen Hawking his voice back.

http://www.nytimes.com/2012/04/03/scien ... .html?_r=2

Mmm… we are not reaching for the stars, we are reaching for other galaxies these days, are we?

Perhaps it’s just me. Silly, pessimistic, glass-always-half-empty me.

The skull and surrounding tissue act not only as a massive attenuator to the EEG signals, but also as a 3-dimensional resistive network that sums and mixes localized EEG sources to that of a fuzzy mess. A network, I may add, that is constantly changing due to instantaneous blood pressure variations and all sort of other parameters.

To try and explain this to a non-technical person, I like comparing it with a very milky, patterned glass that is used in shower doors, etc. If you try to look through that stuff, you will be able to distinguish a cat, medium sized dog, child or adult. But you will have a hard time guessing what type of dog and if it’s a small dog, you will likely mistake it for a cat. You will likely also be unsure if it is a male or female adult standing there. Now try to read the Sunday paper through that stuff!!

So here the guys with optical based DSP algorithm experience will chop in and say – Hang on, but I can increase that signal-to-noise if I know the pattern and characteristics of that glass panel. And yes you can, to a degree. That’s exactly how the Hubble space telescope was fixed, after those dicks cut that primary mirror, or whatever it was, wrong.

The question is this – will you ever be good enough to read the Sunday newspaper through that shower glass? Especially if you only use a low pixel count $2 camera chip! I don’t think so. I am sure that the Hubble telescope guys are still wishing that the optics where manufactured correctly. To read that newspaper, you will need to drill a small hole through that glass panel, and push a pin-hole camera through it.

I think those sort of articles do more harm than good. They over simplify and create over hyped expectations.

I wonder what is in that award winning iBrain device and what sort of electrode it is using. Somehow I think this is just further advertising hype. iBrain? From iRobot?

OK you guys, proof me wrong. Develop an algorithm that can decode what I am thinking now, using a single channel Neurosky or OCZ NIA. No, no, no, wrong. I wasn't thinking of anything that starts with bull.

Regards
Stefan

[yrenard]

Dear Stefan,

thank you for sharing your deception with the OpenViBE community. I totally agree with you on the analysis of how the press relays evolution of science. I had the same feeling about the publication of the video reconstruction stuff a few months ago (could we build up computer pictures of our dream from EEG, really ? ).

Unfortunately, journalists couldn't resist putting science fiction around real science, just to turn it to something even more sensational that it already is. How many times did I read about "reading thoughts" et al ? On the other hand, we should also admit that the messages that get from science/companies to journalists are also usually voluntarily left to interpretation. One did not say anything wrong to the press but just did not exactly describe the appropriate limits to the real stuff.

Having the exact appropriate and honest message, with enough place for perspectives but also with realistic limits is a really challenging task.

Certainly, a single blurry electrode won't be translated to a sound wave form nowadays. It may be used to catch blinks to select and spell letters as with a P300 speller ? Definitely less sexy :p What happened with Stephen Hawking probably stands somewhere between these two options. How far from each ? Only Stephen and people at Neurovigil know

Best regards,
Yann

[Jeff_B]

I don't know much about the SPEARS algorithm, but maybe this technology has more to do with "silent speech" interfaces than BCIs, meaning they don't rely on brain waves but rather on electrophysiological (muscle) signals relevant to the phonatory system (tongue, lip, glottal, larynx,...) ? But I agree this is not what is being said in the media.
Jeff