EEG equipment and signal acquisition Human Cognition and Neural Dynamics Lab Western Washington University
Nov 20, 2014
EEG equipment and signal acquisitionHuman Cognition and Neural Dynamics Lab
Western Washington University
EEG components
Active electrodes
Amplifier and ADC (Analog to Digital Converter)
USB converter
Analog Response Device
Analog Input Box Computer storage and display
Basic Acquisition
• Signals on scalp are very small - microvolt range (1/1,000,000 volts).
• Presents some challenges for acquisition• Acquisition involves– Amplification– Filtering– Digitizing (sampling)– Storage
• Results in one time series per channel (64 in our lab).
Basic Acquisition
• EEG signals are a measure the potential difference between two electrodes.
• Just like the voltage at a battery is the difference between positive and negative poles.
• Thus you always need at least 2 recording electrodes to get a signal.
• In practice we use many electrodes but each EEG signal is always the difference between the signal from 2 or more electrodes.
Electrode placement• Typically adopt an accepted placement scheme for applying electrodes
to the scalp.
• The International 1020 placement system is the most widely adopted.
• It uses a set of measurements relative to landmarks on the head.
• Name reflects the fact that electrodes are placed at intervals that are 10% or 20% of the distance between landmarks.
• Requires distance from front to back of head and distance from left to right.
• Front to back defined as distance from nasion to inion.
• Nasion - intersection of the frontal bone and two nasal bones
• Inion - the most prominent projection of the occipital bone at the posterioinferior (lower rear) part of the skull
Electrode placement
• Requires distance from front to back of head and distance from left to right.
• Left right defined as distance between pre-auricular points.
• Pre-auricular point- root of the zygomatic arch anterior to the tragus
Electrode placement
Electrode placement• Electrode placement begins at 10% from these landmarks.
• Electrodes are placed at 20% intervals.
• Allows for 19 recording electrodes
• Electrode names reflect location.– Even number right/ odd left; z = midline
– C = central; F = frontal; P = parietal; T = temporal; O = occipital
– Larger numbers are farther from the midline
The 10-20 placement system.
Electrode Placement
• Extensions of this placement system include greater numbers of electrodes.
• 10/10 electrode placement places electrodes at 10% intervals.
• 10/5 electrode placement put electrodes at 5% intervals.
• Most labs are using some variant of this system and use the associated electrode names.
The 10-10 placement system
EEG as a time series
• EEG can be considered as a signal that changes over time.
• A simple example is a sine wave that oscillates at a single rate.
• Below are 3 sine waves oscillating at 8 times / sec (Hz).
EEG as a time series
• Waveforms can also be represented in terms of amplitude over frequency –
• And amplitude at different phases.• Can transform data back and forth with no loss of
information.
frequencyphase
EEG as a time series
• EEG is a more complex signal than a simple sine wave
• In theory, any time series – no matter how complex - can be decomposed into individual sine waves of specific frequency and amplitude.
• EEG can be treated in the same way
EEG as a time seriesAmplitude x time Amplitude x frequency
EEG as a time series
• The EEG signal is recorded together with noise that stems from a number of sources.
• Essentially anything that is not the signal of interest is considered noise.
• Noise amplitude is usually larger than the signal of interest.
Sampling theory
• Digital recording of EEG requires sampling brain signals at discrete time points.
• The sample interval (T) is the time between samples expressed in seconds.
• The sample frequency or rate (fs) is the number of samples collected each second expressed in hertz (Hz.)
• fs = 1/T; T = 1/fs
• fs(500 Hz) = T(.002 s)
Sampling Theory
• The sampling theory must be adequate for representing the signal of interest.
• Too low results in aliasing• Too high results in redundancy and unnecessarily
large data files.• If you have to err – always choose to oversample
rather than undersample.• You can always downsample later (lower the
sample rate of the digital signal) but you cannot increase the sample rate of a digital signal.
Sampling Theory8 Hz sine wave sampled at different rates
Sampling Theory
• Nyquist–Shannon sampling theorem: – A signal with maximum frequency f can be
reconstructed using a minimum sampling rate of 1/(2f).
– Given a sampling rate fs, the highest frequency (f) that can be represented is f = fs/2 also known as the Nyquist frequency.
– In practice the sample rate is usually at least 4 times the highest frequency of interest.
Sampling Theory
• Data should not contain frequencies higher than the Nyquist.• Results in aliasing: when a signal appears in the EEG as a
lower frequency
Actual signal (blue) = 20 Hz Undersampling results in aliasing at 2 Hz
Sampling Theory
• Filters must be set to reduce contribution of signal above the Nyquist frequency.– Sample Rate = 250 Hz– Nyquist frequency = 125 Hz– Must low pass filter at 125 Hz.
• High pass filter – allows high frequencies to pass• Low pass filter – allows low frequencies to pass• Notch filter – filters specific range of frequencies• Band Pass – filters all but a range of frequencies
Digital Filtering
frequency
ampl
itud
e
low high
Filtered frequencies
High Pass Filtercutoff frequency
fc frequency
ampl
itud
elow high
Filtered frequencies
LowPass Filtercutoff frequency
fc
Digital Filtering
frequency
ampl
itud
e
low high
Band Pass Filter
fcLOWfrequency
ampl
itud
elow high
Notch Filter
fcHIGH
Pass band
fcLOW fcHIGH
Sources of Noise In EEG
• Capacitive coupling– the electrodes and cables are coupled to signals such as lights,
computers, cell phones, etc. can induce voltage in the leads.
– Theoretically this is the same for all leads so should be removed by common mode rejection.
– In practice, however, this is not always the case so it is best to keep distance between leads and electrical sources.
• Induction– Loop created between body and equipment allows for the formation
of a magnetic field that can induce current flow in wires.
– The best solution is to wrap the cables around each other so that opposing magnetic fields will cancel each other.
Reducing Noise with Biosemi
• Driven Right Left Circuit– Biosemi uses a driven right leg circuit to reduce common mode
signals.
– Uses two electrodes (CMS & DRL) in a feedback loop to drive the voltage of the patient to be the same as the common mode voltage - thereby reducing the effect of external noise.
– CMS used to detect to common mode signal or background noise
– DRL used as part of feedback circuit to eliminate difference between participant and common mode.
– Other systems have only a single ground electrode that grounds the participant for safety reasons.
Reducing Noise with Biosemi
• Active Electrodes– Each electrode has an amplifier attached.
– Amplify recorded signals at the electrode where transduction is occurring.
– Increases size of signal traveling down leads
– Reduces susceptibility to noise in the environment.