From: Doug Sutherland (wearable_at_earthlink.net)
Date: 2002-03-02 03:28:07
For those not on the min-l list, this is from the designer of that
tiny EEG plus L&S unit http://www.pocket-neurobics.com/
Interesting ...
Hi, I'm the manic would developed the Abhayamudra unit, and I
appreciate the time your giving to this.
Looking thru the archives there's a few outstanding issues I should
address, but the main one seems to be a little incredulity that this
little box can do all that is claimed of it. Just a bit of history
then..I'm a satellite comms engineer by training, and after almost 30
years with the same (wonderful) crowd, I had the opportunity to more
on. My wife gave me 6 months to indulge myself & beginning with the
excellent Brainmaster do-it-yourself, two-years later out pops
Abhayamudra. 30,000 lines of hand crafted assembler. (My wife's still
here but only just.)
It does 4ch channel eeg simultaneous - the headset has 6 electrodes (4
active, one reference, one neutral). Ususal positioning would be C3/C4
as front electrodes and P2/P3 or O2/3 for back electrodes, although
frontal is also possible. Each eeg channel can have fixed gain (of
x20,000) or you can switch on auto level control (x10,000..x40,000)
which has the effect of increasing the resolution of the sampling from
10-bit to 12-bit and gets the neurofeedback protocols to work more
effectively. For each eeg channel you can menu-select 35-40-45Hz
bandwidth. And we implement (little talked about) 1/f compensation
(menu selectable) which compensates for the natural roll off in eeg
energy with increasing frequency.
We also include a number of features to try to ensure eeg sessions are
successful (with these big amplifications required, there's lots of
scope for things to go wrong). We have constant monitoring of the
"offset potential" - probably the most common cause of eeg signals
going 'flat-line'. You can measure the impedance between the electrode
and the scalp - as saline evaporates, saline electrodes will increase
in impedance and so will noise. We have what we can an 'electrode
pacifier' (simply passes a few micro-amps thru the electrodes in
saline - place electrodes in the pacifier to 5-10 minutes before use,
and off-set potentials will be less of a problem. (And we provide
instructions on how to make your own electrodes.)
The FFT uses an approximation to a 'hamming' window (thus the use of
the term 'pseudo' in the spec), and it overlaps 4:1. This means the
FFT is done on a block of 64 samples, but after every 16 samples have
been received. This speeds the response time of the FFT, but it is
still slower then digital filtering, so we use FFT for displays &
synchrony analysis, but digital filtering for neurofeedback analysis.
An FFT and three digital filters run concurrently, all the time, on
each of the 4 eeg channels.
Setting the neurofeedback protocol is under menu control. You can have
up-amplitude, down-amplitude, or ratio training individually
selectable for each of the eeg channels - so you can have different
protocols running left-right brain & front-back. You selected one of 9
bands: delta, theta1, theta2, alpha, beta1(smr), beta2, beta3,gamma or
nil as the 'wanted' band and one of the same 9 as the 'unwanted' band.
If both wanted and unwanted are active - this is ratio training:
increasing the amplitude of the wanted to unwanted ratio will generate
'rewards'. If 'unwanted' is 'nil' - this is up-amplitude training; if
'wanted' is 'nil' this is down-amplitude training.
Aural rewards can be chimes (.wav files), chords, or continuously
variable pitch tone (high ratio-> hi pitch).
There's about 15 displays which vary from the 'dance of the dendrites'
in the frequency domain (FFT readout) to games such as 'Pong' (what we
call Halley's Comet) where the size of the bats are controlled by the
ratio or amplitude on each of the 4 eeg channels. (One channel
controls one edge of one bat.)
Light & Sound is conventional binaural beat and flashing glasses. We
have a "sequencer" which allows to to overide the preset (I should
have said there are about 10 preset configurations which set defaults
for neurofeedback & light & sound, plus 3 which the user can program).
In 'mixed' mode, the binaural beat & the light-glasses flash rate is
modualted by the eeg signal left & right brain separately. A switch on
the glasses allows left eye - right eye or left field of vision/right
fov.
Binocular rivalry (Michael O'Bannon is the expert here) is not quite
the same as eye dominance, but probably related. If sufficiently
differentiated images are presented separately to the left & the right
eye, you will not see a merging of the images but rather one or the
other, and the images you see will alternate at a quite regular
periodicity - the 'switching rate' (of about 1 sec - unless you are M
O'B where it will be in excess of 2Hz rate). So what? A search on
"Pettigrew binocular rivalry" will explain his research finding that
manic depressives are likely to have slow switching rates. No evidence
that this switching rate can be trained, so we simply include it a s a
monitor, and you can track your switching rate over time, mode, etc.
We supply a PVC "stereoscope" to isolate left & right eye images.
The unit allows you to graphically trace ratio/amplitude over the
previous 30 minutes, as well as logging to non-volatile memory to
track progress over a longer period. A real-time clock allows logged
data to be date & time-stamped, and alarm allows the unit to
automatically turn itself off (at night) or on (in the morning).
The infrared link allows communication to another Abhayamudra II (they
can exchange 4ch eeg raw data over the link) to mix your eeg signal
with that of your partner.
We supply a small IR adaptor that you can velcro to the side of your
monitor. The 4ch raw eeg data can be streamed to the PC, and the PC
can command the unit. If the program on the PC were capable, this
mixing of eeg signals can be extended over the internet to remotely
located Abhayamudra units. >>>The PC application to do this does not
yet exist<<<< but we do supply the IR adaptor, and most General
Purpose comms programs can accept the streamed data and file it (we
recommend shareware "SerialWatcher"). We hope to have at least
rudimentary software by mid-year, and would be keen to hear from
individuals who may have software they would like to adapt.
Apologies for the length of reply and its 'infomercial' nature, but I
think its necesssary at this point.
Happy to take further queries offline.
Bruce McMillan
bruce.mcmillan_at_Pocket-Neurobics.com
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:39 BST