I've been thinking about modeling the EEG measurement and how to write up the corresponding chapter. I think this is how it'll go. First, I will introduce the concept of impedance in a circuit element.
Impedance is the relation between voltage and current for a two terminal circuit element. In other words, if I give the element a voltage V, what is the resulting current I? The relation, impedance Z, can be expressed as a complex number. This complex notation allows us to describe responses over time. Except in ideal elements, giving an element voltage V will result in a current which monotonically and eventually, so not instantaneously, moves to a final current I.
Perfect, from there I can give a simple model of the voltage fluctuations at the scalp: a voltage source, with impedances Z1 and Z2 on each side. Of course, in real life it is more complex, the number of sources is much greater than 1, each with its own Z1 and Z2. Between the sources and the surface of the scalp, there are other sources - some of which may be more stationary, and some of which may be more dynamic. It is known, for instance, that across the skin barrier there is effectively a voltage source in the 10s of mV. This is related to a difference in ionic concentration in the outer surface and the inner surface of skin. When the subject perspires, pores are opened and filled with sweat which then changes the ionic concentration. Fortunately these changes happen rather slow, and are only relevant to recordings at very low frequencies.
We can begin our analysis with the simple model of a single source and Z1 and Z2.
Impedance of skin.This should be a good sized section, because it is important. First I will review different measures - in electrophysiology, for EIT, and in dermatology, to test the results of various cremes, etc. I should talk about the technique, and then look at the results.
For some time now I have been interested in developing some browser toolbars, though I have yet to actually sit down and do one. On the way though, I've run into some good references, one of which is the MSDN reference on browser extensions. Visual Basic Shell Programming by J. P. Hamilton is also quite good, and probably would be my recommended approach. You can snag the code examples from the O'Reilly site.
I'm currently using a CS5532-BS to measure noise from dry EEG sensors. After some googlin' this afternoon and browsing through datasheets, it dawned on me that there isn't really a straight-forward way of comparing ADCs. It would be nice to have a website where one enters in the frequency range of interest and the voltage range, and then is given a plot of noise spectrum for various ADCs. You could also have a setup where you can add a number of different front-end amplifiers and see the final result. Alternative vendors include Linear Technology and TI. I've seen the LTC240x and the ADS121x mentioned on Google Groups though I'm not sure how up to date that is. The key, of course, is simply to have noise that is below that of your front-end. Having noise much much smaller really doesn't make that much of a difference.