|
Electroencephalogram (EEG) is a device that amplifies and displays brain electrical potentials.
Every EEG consists of brain electrical potentials amplifier, digitalization (conversion of analog signal to ones and zeros – binary numbers) and displaying of data. Based on this model, and some useful advices from openEEG project I developed an EEG system, let’s analyze each part separately (to completely understand this section you need some basic electrical knowledge).
EEG amplifier
EEG amplifier, which block schematic is shown on picture, gets input signal directly from an examinee head.
EEG - block shematic
Amplifier is connected with head using electrodes (peace of wire with silver-chlorine, AgCl, plate at one of the wire ends with radius of about one centimeter). We need one electrode to get general information about noise so we can do some more quality amplification of signal. To collect the signal we need two electrodes for each channel. Electrode placing should be done using the 10/20 system, explained in article about brain electrical potentials.
Amplitude of brain potentials measured directly on a scalp is about 100uV, and its frequency range is not strictly defined, we can say that typically most of spectrum energy is between 0.2Hz and 20Hz.
Beside the scalp potential, on amplifier entrance there is a polarization voltage (connection between scalp and electrode plate becomes a little battery with voltage of about 100mV, that voltage is called polarization voltage, it’s always DC voltage – it makes it easy to remove it later in circuit).
On the amplifier entrance there is also a noise that is inducted in electrode wires (especially from electrical wires – 50/60Hz, but also from any other electromagnetic source). So first part of EEG amplifier is a passive RC filter that filters high frequency components.
Next part is protection circuit that protects both, examinee from short circuit in device, and the device from connecting it to some power source instead of to brain.
After protection circuit there is an instrumentation amplifier. Instrumentation amplifier is amplifier that amplifies signal difference and rejects input signals common to both input leads. This is very important cause noise is pretty same on both instrumentation amplifier input leads (electrodes and electrode cables are very close so noise influence is same on both of them), so due to its ability to reject input signals common to both input it will reject the noise. Brain potentials are different on each electrode so instrumentation amplifier will amplify brain potentials because it amplifies signals difference.
I used instrumentation amplifier INA118, very quality and suitable for this application that rejects common signal with 80dB (common mode rejection ration, CMRR = 80dB), has very low offset voltage of 50uV (DC component that amplifier adds to the signal), very low drift voltage of 0.5uV/°C (another DC component that amplifier adds to the signal that is a function of temperature). I set the instrumentation amplifier to amplify 12 times; it can not be set to amplify too much because of DC component (power supply is +/- 5V and amplified DC component can’t exceed power supply because in that case there would be no “place” for useful signal).
Next step is removing of the DC component with passive high pass filter with cutoff frequency of 0.15Hz.
After filtering, signal is amplifier with high precision, low power operational amplifier TLE2022 for 1 – 200 times (there is a switch for choosing amplification level).
Amplification brings some new DC component to signal cause of amplifier offset and drift and it’s removed with another passive high pass filter with cutoff frequency of 0.15Hz.
Before digitalization active fourth row low pass filter will cutoff all frequency component higher then 30Hz cause they are not very interesting for EEG analysis (they are very small) and we need to limit highest frequency component to be at leas twice smaller then sampling frequency to prevent spectrum overlapping known as aliasing. Designed active filter beside filtering also amplifies signal for 10 times what in total means that signal can bi amplified from 120 to 24000 times.
Digitalization of amplified brain potentials
Amplified brain potentials we need to digitalize - convert from analog, continuous signal, to discrete signal. We do both, amplitude, and time quantization (digitalization).
This system uses analog to digital converter ADS7825 for quantization with sampling frequency of 200Hz and amplitude resolution of 16 bits.
Sampling frequency of 200Hz means that system takes one sample of signal every 1/200 = 0.05s, or 200 samples in second.
Amplitude resolution of 16 bits means that we will divide input dynamics of our analog to digital converter (20V because its input range is +/-10V what means that difference between maximal and minimal signal is 20V what is definition of input dynamic) with 65536 (because 16 bits resolution means that we have 2pow16 = 65536 discrete amplitude levels) to get an elementary amplitude level (also know as level of quantization noise). In dependence how big is our signal it will get one of discrete amplitude levels. For example if signal on the analog to digital converter input is -10V it will get amplitude level 0, if the signals amplitude is 0V it will get 65536/2 = 32768 amplitude level and if the signal is 10V it will get 65536, the maximal amplitude level.
ADS7825 has four analog to digital conversion channels so it is possible to connect four EEG amplifiers.
After digitalization, system uses ATmega8 microcontroller to collect data from analog to digital converter, to process them (remove the 50Hz/60Hz noise produced by electrical wires using a finite impulse response (FIR) filter) and send them using serial RS232 connection to PC. ATmega8 has six 10bits analog to digital converters, the system uses two of them for stimulus (signals that can be used for brain stimulation during EEG recording) acquisition.
Before sending signal to PC due to security reasons we need to isolate the device from a PC. Isolation is done using CNY65 opt couplers, devices that transform small signals from the device to a PC but for a big signal they are great isolators – they can handle voltage of 8kV for 60s, so in case of any damage of PC (for example from lightning) the examinee will be safe.
The device is connected with a PC using a RS232 cable, but it is possible to use RS232/USB converter and connect the device to PCs that don’t have RS232 connector.
PC application for saving and displaying data
The application is developed in Visual studio 6.0, it collets data that comes to serial RS232 port or to USB port (if RS232/USB adapter is used) displays them on screen and saves to a file. User can choose how many EEG (1 to 4) and stimulus channels (0 to 2) he wants to watch and save.
The file that the application creates can be read with MATLAB (the best signal processing software). The application also creates an information file which consists of examinee name, date and recording description.
The Application works under Windows XP, Windows Millennium, and Windows 2000 operating systems.
EEG application with test, 18Hz impunt signal
Power supply
As special power supply is very important to EEG system so I will say few words about it.
The whole system gets power supply from PCs USB port.
The power source is isolated by using DCP010507DB DC/DC converter that converts 5V voltage from USB to +/-7V. It is necessary to have such isolation part to prevent examinee injury in case of big electrical network damages. Converted voltage is filtered using so called “pi filter” to reject DC/DC converter rippling. After filtering voltage is stabilized using 7805 and 7905 stabilizers. That stabilized voltage supplies the whole system (digital and analog part).
Except of this design I am planning to build few more, very similar, but simpler EEG devices.
One variant will be using the microcontroller’s analog to digital converters instead of ADS7825. The design is then much simpler and cheaper but the amplitude resolution is worse then with ADS7825 but it’s still good enough to see nice EEG.
Other variant is to use the device just with portable PC when it works on batteries (when all connections with electrical network are switched off the computer). This allows removing all isolations in circuit (CNY65 and DC/DC converter) because the PC is not connected to electrical network and then there is no need to protect from it.
Take a look at EEG implementations in self hypnosis, light brain stimulator and sound brain stimulator
|
