In <95072.212337IO20763 at MAINE.MAINE.EDU> <IO20763 at MAINE.MAINE.EDU> writes:
>Hello,
>I have a question for anyone who has analyzed electrophysiological information
>with power density spectra analysis.
I am using it to analyze electroretinograms, both oscillatory potentials
and in comparison with stimulus frequency. Just started doing this, and
have not published anything of it yet, but the reasons for using it is to
find the dominant frequency as well as second harmonics in the response.
it is tricky; I am not yet confident with it, and keep analysing signals
with a known frequency just to convince myself!
> I have recently conducted hundreds of ele
>ctrophysiology experiments on the olfactory epithelium of American eels. I rec
>orded oscillatory potentials of many different chemicals. I wish to use PDS an
>alysis but would like to know how the responses should be quantitatively presen
>ted.
Well, the convention in the ERG literature is to show a typical signal
(at least one!) in the time domain, and then power spectra of signals,
i.e microWatts/Hz as a function of frequency. Some people also use
bandpass filtering before analysis and then the resultant spectra.
Various components of the ERG with a low frequency (a- and b-waves) have
a high power, while high frequency components like the OPs have a low
power when the signal is unfiltered. But these have a different
physiological basis. Also, the length of the analysed window in the time
domain has an effect on the spectrum (of course) so people sometimes do a
number of spectra on many windows taken after the stimulus. I would think
this might be important in the EOG also?
> It seems that presenting the frequencies of the responses will not make t
>hem comparable to measurements made with more conventional amplitude measuremen
>ts. Should the frequencies as well as amplitudes (voltages) be presented or is
> it now common practice to only present frequency data?
Don«t know about EOGs, but the oscillatory potentials of the ERG have a
dominant frequency of about 140Hz while they are "riding on" on the
b-wave with a dominant frequency of 6 Hz. You want to separate these
clearly and measure them in relative terms, so Fourier analysis can give
you info that conventional amplitude measurement can not do. Do the OPs
in the EOG have a physiological significance beyond the slow components?
I am wondering because there is a guy in my dept. actually doing EOG in
fishes, and i do not think he uses FFT, just amplitude measures (could be
wrong). In the ERG literature people like to see at least on signal in
the time domain, but there are "frequency freaks" who only present power
spectra! Usually it has a sound physiological basis though, and provides
more information, for instance if the stimulus has greater or smaller
effects on high or low frequency components etc. From pwoer density
spectra you can see this more easily than from the signal in the time domain.
Thor Eysteinsson
Dept. of Physiology
University of Iceland
