Specifying hardware for an EEG project (not a homework/software question)

[wnd3177]

I've been looking at the basics of EEG DAQ and am wondering what NI hardware is suitable for such a task and what specs are the most important to pay attention to.  I'm trying to learn more about specing out DAQ and signal conditioning hardware here more than I am actually trying to build an EEG machine so I don't see it worth calling NI and speaking with sales / applications engineers regarding this.  

I believe EEG range from 0.001 mV to 1 mV and the electrodes must be isolated from mains.   A few questions...  

If I already have DAQ cards of sufficient bandwidth and resolution would I need only an instrumentation amplifier?  

I've read that 2,000x is the typical max amplification in signal conditioning.  Would the amplifier be responsible for keeping noise out of the system or would a special DAQ card be required as well, with voltages in the 2 mV to 2 V range following amplification?  

If a signal (any signal) required more than 2,000 amplification, would two instrumentation amplifiers be used in series?  

How stringent are these specs?  (i.e. Would I need the 'top of the line' amp or would several amps meet these requirements) ?  

Does NI manufacture DAQ devices that would meet these requirements alone such that an amp would not be required?  

What other hardware considerations might I need to make?  

I took the Data Acquisition and Signal Conditioning class offered by NI but there wasn't too much focus on low-level signals from what I remember.  Thermocouples and strain gauges are the topics I remember studying.  

Any information would be great.  Thanks!   

Message Edited by wnd3177 on 08-20-2008 07:24 PM

[Steve.J]

Hi there,

I'm an NI employee but I have a history as a former medical device engineer and actually have a background in EEG and other biosignal acquisition/processing.  Your asking lots of good questions!

I'll try to answer them in order:

I believe EEG range from 0.001 mV to 1 mV and the electrodes must be isolated from mains.   A few questions...  

If I already have DAQ cards of sufficient bandwidth and resolution would I need only an instrumentation amplifier?  

No - you need also an isolation amplifier or some other way to provide protection from electrical shock, was well as isolation from "AC mains" (line voltage) if the system is powered off the wall.

I've read that 2,000x is the typical max amplification in signal conditioning.  Would the amplifier be responsible for keeping noise out of the system or would a special DAQ card be required as well, with voltages in the 2 mV to 2 V range following amplification?  

You don't need a special DAQ card if the signal conditioning system is properly designed.  The whole trick is to keep noise from ever getting digitized in the first place using proper filtering and really good common-mode rejection (CMRR) - the main job of the first differential amplifier.

If a signal (any signal) required more than 2,000 amplification, would two instrumentation amplifiers be used in series?  

Yes - typically it is difficult to get that much gain in one stage and still keep the other specs that you need.  Most bioamplifers are AC coupled to reject the large electrode offset potential generated by the electrode/electrolyte/skin interface.  This means that your 50 microvolt EEG signal is riding on top of a 100mV DC offset that shifts around with movement and time.  You need to block this offset or it gets amplified (it is a differential DC signal) and saturates your amplifier, so most of the amplification happens after the AC coupling in a second stage.

How stringent are these specs?  (i.e. Would I need the 'top of the line' amp or would several amps meet these requirements) ?  

The specs aren't that stringent since typically EEGers don't care much about absolute measurement accuracy - there more concerned about fidelity in frequency and relative amplitude.

Does NI manufacture DAQ devices that would meet these requirements alone such that an amp would not be required?  

No.  

What other hardware considerations might I need to make? 

Many safety and other considerations.  It's definitely not a trivial design problem.  I would recommend finding some good bioamplifiers and concentrate on the rest of the data acquisition problems, unless the signal conditioning is the whole point.  You can even find old EEG machines on Ebay (like this one:  http://cgi.ebay.com/Grass-Instruments-Co-8-10B-Electroencephalograph_W0QQitemZ250180877963QQcmdZViewItem?_trksid=p3286.m20.l1116 ) that have amplifier output jacks that can go straight to a DAQ card).

Good luck!

Steve 

 

[G-Mann]

Namaste 🙂 !!!

Needed help regarding the design of a simple (not robust and complicated) EEG signal acquisition device...

I needed to know what kind of hardware id necessary to just isolate a Certain freq component and use that (mainly the amplitude) to drive a Fan...

A little heads up would reallly help my dire confusion!!! 😞

Thanks

Grishma 

[Steve.J]

For what you want to do, all you need is a basic bioamplifer (battery powered, isolated EEG differential amplifier) with fixed gain and bandwidth (something like 1000x gain and 0.5 to 30Hz) going to a low-cost USB data acquisition device (like the NI USB-6008) and digitize at something like 200s/sec.  You can do the power spectra measurements in LabVIEW to isolate Alpha, Delta, Theta, etc. frequencies and determine power in each band.  The output of these measurements can be used to modulate an analog output (or alternatively a digital PWM signal) from the USB-6008 to control the power to your fan.

Steve

[G-Mann]

Thanks for the reply...The tips helped...But is there a way of doing this without a PC?

Grishma 

[Steve.J]

A PC is the quickest/easiest way to do it, but you could create a processing board with a DSP or microcontroller on it to do the FFTs of the EEG signal (spectral measurements).  You could also try to create narrow bandpass filters in the analog domain to extract just the frequency bands of interest, then rectify/integrate the outputs to get a general "power in band" kind of output - all in analog circuitry.  These days, most people get to digital as fast as possible so that they can do all the signal processing in the digital domain which dramatically increases your options for signal processing, all at very low cost and power.  You can get cheap DSP chips with A/D converters built into them. 

Actually - NI makes a small DSP board that you can target with a add-on module (The LabVIEW DSP Module for Speedy-33).  This is sold mostly to the academic community for teaching/demonstrating signal processing, but it might be a perfect little board for doing what you need to do.

Good luck!

Steve