From: Doug Sutherland (wearable_at_earthlink.net)
Date: 2002-03-06 05:30:15
Jim,
> BWView was always intended as an analysis app. Another approach
> to generating the analysis would definitely be needed for real-
> time use and biofeedback (maybe IIR filters, or convolution the
> direct way (FIR filters) using a smaller set of selected
> frequencies).
Excellent point. For biofeedback, the speed of actual feedback
to human is important. FFT is slow in this respect relative to
digital filtering (IIR or FIR). Brainmaster uses a butterworth
low-pass digital filter in software. Here is Tom Cullura's
brief description of the quadrature digital filtering that is
used in Brainmaster.
http://www.biof.com/brainmastersoftware/quad1.doc
The BrainMaster uses quadrature filters for all internal filter
operations. The low-pass filter is a 6th-order Butterworth
filter, realized as an IIR filter. This is a very stable filter,
with minimal overshoot, and excellent transient response. This
filter provides a rolloff of 36 dB per octave, based on the 1/2
width of the filter. For example, if a filter is set to pass
12-15 Hz, the low-pass filters will be 1.5 Hz wide, with a
center frequency of 13.5 Hz. Therefore, the -3dB points will be
at 12 and 15 Hz. One octave beyond that will be 3 Hz away from
the center. Therefore, at 10.5Hz and 16.5 Hz, the response will
be -36dB down, a factor of nearly 100.
More Collura on filtering ...
During this assessment, real-time display of EEG frequencies
is desirable. This may be achieved either with a concurrent
spectral (FFT) display, or with the use of digital filters,
so that each frequency band occupies its own waveform display
channel. Two methods are available for the measurement and
display of EEG frequency bands: Spectral display, generally
using the Fast Fourier Transform (FFT), and digital filtering,
which provides a real-time waveform display limited to the
frequencies of interest. It is generally accepted that the
use of digital filtering is superior, in that it provides a
more rapid indication of changes. If the feedback is to be
useful to the patient, it must provide a rapid and accurate
indication of the frequency content of the EEG. Generally, a
response time of 250 milliseconds is considered the maximum
tolerable, and response within one or two cycles of the
pertinent rhythm is preferred. In the case of a 15 Hz rhythm,
this would be 1/15 second, or approximately 75 milliseconds.
>From http://www.hackcanada.com/ice3/wetware/grounds.html
This is an interesting article "Human Steady-State Visual and
Auditory Evoked Potential Components During a Selective
Discrimination Task" from the Journal of Neurotherapy ...
http://www.snr-jnt.org/JournalNT/JNT(1-3)1.html
Several techniques are available for measuring evoked responses
in a more rapid manner (McGillem & Aunon, 1977; Vaz &Thakor,
1989). One which is well suited to real-time brain monitoring
is synchronous filtering of steady-state evoked potentials
(Collura, 1990).
And this from a seller's web site ...
The Brainmaster has digital filtering to show the frequency
make-up of any signal between 1 & 40 Hz. A Mind Mirror type
of display can be viewed as well as a Compressed Spectral
Display (CSD) that shows a 3D picture of the spectral data.
The CSD lays down the latest brainwave spectral activity every
three seconds gradually forming a five minute "picture" of
brain activity. This type of spectral processing is also used
for the various games such as Pac Man, Space Rocket and Happy
Face.
>From http://www.comptronic.com/brainmtr.htm
Some relevant resources on digital filtering ...
Real Time Feedback: Is it real?
http://www.focused-technology.com/realtime.htm
CMSA Filter Designer
http://www.cmsa.wmin.ac.uk/filter_design.html
Filtering Classes and Functions
http://spuc.sourceforge.net/filter.html
MathTools Java Filtering
http://www.mathtools.net/Java/Filtering/
This EEG Viewer uses IIR Filtering
http://www.cn.stir.ac.uk/~bp1/eegviewer/
Some info on time/frequency EEG analysis (wavelets)
http://brain.fuw.edu.pl/~durka/tf/tf.html
http://bigwww.epfl.ch/publications/schiff9401.html
http://dol.uni-leipzig.de/pub/1997-16/en
http://www.ecse.rpi.edu/Homepages/das/eeg.html
http://complex.gmu.edu/neural/abstracts/wavelet.html
More wavelet info
http://www.public.iastate.edu/~rpolikar/WAVELETS/WTpreface.html
http://www.public.iastate.edu/~rpolikar/WAVELETS/waveletindex.html
http://www.cosy.sbg.ac.at/~uhl/wav.html
http://www.amara.com/current/wavelet.html
Namaste,
Doug
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:39 BST