Biosignal Processing

EEG Signal Processing

Biomedical Engineering Department

JiT, Jimma University

Gizeaddis L. (MSc.)
 Electroencephalogram (EEG)

• EEG measures the activity of large numbers (populations) of

neurons.

• EEG recordings are noninvasive and relatively low-cost.

• Electrodes measure voltage-differences at the scalp in the

microvolt (μV) range.

• Voltage-traces are recorded with millisecond resolution –

great advantage over brain imaging
 Electroencephalogram (EEG)

• Brain Activity

1. Spontaneous activity

• is measured on the scalp or on the brain

• The amplitude of the EEG is about 100 µV when

measured on the scalp, and about 1-2 mV when
measured on the surface of the brain.

• The bandwidth of this signal is from under 1 Hz to

about 50 Hz,
 Electroencephalogram (EEG)

• Brain Activity

2. Evoked potentials

• are those components of the EEG that arise in response

to a stimulus (which may be electric, auditory, visual, etc.)

• Such signals are usually below the noise level and thus not
readily distinguished, and one must use a train of stimuli
and signal averaging to improve the signal-to-noise ratio.
 Electroencephalogram (EEG)

• Brain Activity

3. Single-neuron

• Single neuron behavior can be examined through the

use of microelectrodes which impale the cells of
interest.

• Through studies of the single cell, one hopes to build

models of cell networks that will reflect actual tissue
properties.
 Basic Acquisition
• Signal can be acquired from the scalp, surface of the brain
(Electrocorticography) or directly from the neuron by using
microelectrodes.
 Basic Acquisition
• EEG signals are a measure of the potential difference
between two electrodes.

• Thus you always need at least 2 recording electrodes to get a

signal.

• In practice many electrodes are used but each EEG signal is

always the difference between the signal from 2 or more
electrodes.
Basic Acquisition of EEG
 Basic Acquisition of EEG

Many neurons need to sum their activity in order to be detected by EEG electrodes. The
timing of their activity is crucial. Synchronized neural activity produces larger signals.
Electrode placement

Standard placements of electrodes on the human scalp:

• A, auricle;
• C, central;
• F, frontal;
• Fp, frontal pole;
• O, occipital;
• P, parietal;
• T, temporal.
 Electrode placement
Nasion
• The electrodes are placed in
equal intervals.

• 10-10 electrode placement

places electrodes at 10%
intervals.

• 10-5 electrode placement put

electrodes at 5% intervals.

• 10-20 is widely adopted Inion

The 10-10 placement system
 Electrode/Lead Type
• Bipolar or unipolar electrodes can
be used in the EEG measurement.
– Bipolar: potential difference between
a pair of electrodes is measured.

– Unipolar: the potential of each

electrode is compared either to a
neutral electrode or to the average
of all electrodes
EEG
 EEG Rhythms

• EEG potentials are good indicators of global brain state.

• EEG rhythms correlate with patterns of behavior (level of
attentiveness, sleeping, waking, seizures, coma).
• They often display rhythmic patterns at characteristic
frequencies
• it is possible to differentiate alpha (α), beta (β), delta (δ),
and theta (Θ) waves as well as spikes associated with
epilepsy by their frequency.
 EEG Rhythms
• General patterns of Behavior
 EEG Rhythms
• Rhythms occur in distinct frequency ranges:
 EEG Rhythms

• Higher frequencies:

o active processing, relatively de-synchronized activity (alert

wakefulness, dream sleep).

• Lower frequencies:

o strongly synchronized activity (non dreaming sleep, coma).

 The Basic Principles Of EEG Diagnosis

• The EEG signal is closely related to the level of consciousness

of the person

– As the activity increases,

• the EEG shifts to higher dominating frequency and lower amplitude.

– When the eyes are closed,

• the alpha waves begin to dominate the EEG.

– When the person falls asleep,

• the dominant EEG frequency decreases.
 The Basic Principles Of EEG Diagnosis
• The level of consciousness in the EEG signal
 Determining waveforms from EEG Signal
Power spectrum:
• Plot of the portion of a signal's power (energy per unit time) falling within
given frequency bins
 Determining waveforms from EEG Signal
ERP
ERP’s are obtained after averaging EEG signals obtained over
multiple trials (trials are aligned by stimulus onset).
 Applications of EEG

• Psychological Research

• Neurological Research

• Medical Research

• Educational Research and Application

• Therapeutic and Monitoring Application

 Clinical Applications of EEG

• Main application areas of EEG

– Epilepsy

– Emotion study

– Sleep Disorder studies

– BCI (research oriented perspective)

 Clinical Applications of EEG

• Epilepsy
– A disorder of the central nervous system characterized by
loss of consciousness and convulsions
– A person with epilepsy suffers from seizures during which
sudden bursts of uncontrolled electrical activity occur in a
group of neurons of the cerebral cortex.
– classifying seizures into different groups has been
established based on the characteristics of the EEG
 Clinical Applications of EEG
• Epilepsy
– The EEG is the principal test for diagnosing epilepsy and
gathering information about the type and location of seizures.
 “Right and Left Brain” Characteristics

• Right Brain:
• Spatial processing

• Musical tasks

• Left Brain:
• Verbal processing

• Mathematical skills

** Emotions have also been correlated with differential hemispheric

processing (Davidson, Schwartz, Saron, Bennett and Goleman, 1979)
 Hypnosis, Imagery, Meditation and
Perception
• Traditionally, the lines between the four greatly related
mind states were not very clear.

• Research conducted with the use of EEG, however, has

enabled researchers to draw more distinct lines between
each of the topics, and to study the distinct
characteristics of each of them.
 Sleep and EEG:

• Benefits of sleep studies:

- Better understanding in studies involving
A. Human performance
B. Behavior
C. Well-Being
• Sleep disturbances are the second most frequent indicator of
illness.
 Sleep and EEG:

• Sleep staging is crucial for the diagnosis and treatment of

sleep disorders.

• It also relates closely to the study of brain function.

– In an intensive care unit, for example, EEG wave classification is used
to continuously monitor patients’ brain activities.

– For newborn infants at risk of developmental disabilities, sleep

staging is used to assess brain maturation.
 Sleep and EEG:
• Sleep studies began in the 1800’s

• 1930’s EEG recording machines began to make an appearance.

• Different kinds of physiological machines and recorders

– EEG

– EOG (electroculogram)

– EMG (electromyogram)

– Rectal temperature

– Respiration
 Sleep and EEG:

• Setup for acquisition of a polysomnographic recording in the

sleep laboratory.
 Sleep and EEG:

• Different stages of sleep and their respective brain waves:

– Stage 1: Low voltage random EEG activity (2-7 Hz)

– Stage 2: Irregular EEG pattern/negative-positive spikes

(12- to 14 Hz)
– Also characterized with sleep spindle and K-complexes that
could occur every few seconds.

– Stage 3: low/high voltage waves and high amplitude delta

waves or slow waves (2 Hz or less).

– Stage 4: Delta waves

 Sleep and EEG:
– Stage REM (Rapid eye Movement): “episodic rapid eye movements,”
low voltage activity.

– Stage NREM: All stage combined, but not including REM or stages
that may contain REM.

• The K-complex occurs randomly in stage 2 and stage 3

– The K complex is like an awaken state of mind in that is associated
with a response to a stimulus that one would experience while awake.
 Sleep and EEG:

• Light sleep (stages 1&2)

• Deep sleep (stages 3&4)

• Sleep is cyclical

• Meaning that one will usually go from light sleep to

deep sleep back to light sleep again.

• The whole cycle take about 1 and ½ hours (90 min.)

EEG brain waves in the Sleep Cycle:
 EEG and Dreaming:

• REM was observed as fast eye movements that moved in

many directions while a person was asleep.

• Studies showed that people remember dreams 75% (60-90%)

more when waken during REM sleep,

• If not woken during REM sleep dreams are only remembered

7% of the time.

• NREM dreams are described as being less active

 EEG for Dream Studies:
• The dream studies in:

– Changing dream content

– Drugs and their effects on dreaming

– Pre-sleep stimulation and dreaming

– Dream content with respect to patients that had different

psychiatric disorders
 Sleep Deprivation
• Three different kinds: (Naitoh, 1975)
– Total sleep deprivation

• missing one or more sleep periods

– Partial sleep Deprivation

• missing a section of the sleep cycle

– Differential Sleep Deprivation

• Wakening a person during different random points in a sleep cycle
based on EEG signs of particular stage
 EEG for BCI
• Direct (non-muscular) communication channel between the
brain and the external world
EEG Signal Processing
 EEG Artifacts
• One of the biggest challenges in using EEG is
– the very small signal-to-noise ratio of the brain signals
– coupling by the wide variety of noise sources.

• The artifacts in the EEG signal appear during the acquisition due
to different causes,
– bad electrodes location,
– electrodes impedance,
– bioelectrical signals from other parts of the body (heart and
muscle activity,
– eye blink and eyeball movement.
 EEG Artifacts
• Generally Artifacts can be categorized into
– Physiological origin
• Eye movement and blink

• Cardiac activity

• Muscle activity

– Technical origin
• Electrodes

• Equipment interference
 Eye Movement Artifacts
• Eye movement produces electrical activity- the
electrooculogram (EOG)--which is strong enough
to be clearly visible in the EEG

• The strength of the interfering EOG signal

depends on
– the proximity of the electrode to the eye and

– the direction in which the eye is moving, i.e., whether

a horizontal or vertical eye movement takes place.
 Eye Movement Artifacts

• Artifacts in the EEG caused by (a) eye movement and

(b) repetitive, voluntary blinking
 Eye Movement Artifacts
• From an artifact processing viewpoint, it is highly
practical if a "pure" EOG signal can be acquired by
means of two reference electrodes positioned near the
eye, which do not contain any EEG activity.

• They can be used as a reference signals which are useful

for artifact cancellation purposes
 Artifact by Muscle Activity
• This artifact is caused by electrical activity of
contracting muscles, measured on the body surface
by the EMG.

• This type of artifact is primarily encountered when

the patient is awake and occurs during swallowing,
chewing, talking, sucking, and hiccupping

• The overall shape of the EMG signal depends on the

degree of muscle contraction:
 Artifact by Muscle Activity
• a weak contraction produces a train of low- amplitude spikes,
while an increase in contraction strength decreases the inter-
spike distance so that the EMG more closely exhibits the
properties of a continuously varying signal

• They overlap with beta activity in the 15-30 Hz range.

 Artifact by Cardiac Activity
• The electrical activity of the heart, as reflected by the
ECG, can interfere with the EEG.

• Although the amplitude of the cardiac activity is usually

low on the scalp in comparison to the EEG amplitude it
can hamper the EEG considerably at certain electrode
positions and for certain body shapes, e.g., subjects with
short, thick necks.
 Artifact by Cardiac Activity
• The repetitive, regularly occurring waveform pattern which
characterizes the normal heartbeats fortunately helps to
reveal the presence of this artifact

• The spike-shaped ECG waveforms can sometimes be

mistaken for epileptic activity when the ECG is barely visible
in the EEG

• ECG can be acquired independently by one or several

electrodes for use in canceling the ECG activity that may
be superimposed on the EEG.
 Power line Noise

• EEG signals, are often contaminated with the 50/60 Hz line

frequency interference from wires, other equipments which
are captured by the electrodes and acquisition system.

• These causes artificial spikes in the EEG.

• They are distributed in several channels of EEG and can

made a mistake in the analysis of the record.
 Baseline Noise
• Poor contact of the electrodes and perspiration of the
patient under the electrodes may affect the electrode
impedance which causes low frequency artifacts.

• Baseline drift may sometimes be caused by variations in

temperature and bias in the instrumentation and amplifiers
as well.

• This type of noise is undesired and needs to be

removed before any further signal processing, for proper
analysis and display of the EEG signal.
 Artifact processing
• Artifact processing is synonymous with a preprocessing

stage which conditions the EEG signal for subsequent

analysis.

• The common approach to artifact processing is to

– first estimate the noise v(n), either from a signal measured on

the scalp or from available reference signals, and

– then to subtract the estimate from the observed signal x(n).

𝑺(𝒏) = 𝒙(𝒏) − 𝒗(𝒏)

 Artifact processing
• Another type of model may be preferred which assumes

that the signal and noise interact in a multiplicative way,

𝒙 𝒏 = 𝑺(𝒏)𝒗(𝒏)

• This noises are dependent on the system (signal)

• Once the noise is modeled the next step is filtering

 Artifact Reduction Using Linear Filtering

• Removal of 50/60 Hz power line interference can be

done with a linear, time-invariant notch filter.

• Linear, time-invariant filtering (Low pass filtering) has

been considered for the reduction of EMG artifacts

– Best only when analysis of slower EEG rhythms is of particular

interest
Artifact Reduction Using Linear Filtering

EEG signal before and after 50/60 Hz power line

interference removal
 Artifact Cancellation Using Linearly
Combined Reference Signals

• Since artifacts due to eye movement and blinks are very

common, most efforts have been directed toward

developing cancellation methods for those artifacts.

• The EOG artifact is first computed, based on one or

several reference signals ("EOG-only" signals), and then

subtracted from the EEG signal measured on the scalp

 Artifact Cancellation Using Linearly
Combined Reference Signals

• We assume that the EEG signal is composed of cerebral

activity s(n) which is additively disturbed by the EOG

artifact v0(n),

• To produce an artifact-cancelled signal S(n) by subtracting a

linear combination of the reference signals from the EEG,

using the weights

 Artifact Cancellation Using Linearly
Combined Reference Signals

• Electrode positions for the recording of EOG signals which

reflect horizontal (F7- F8) and vertical (Fp2- I2 or Fpl- I1) eye

movement
Artifact Cancellation Using Linearly
Combined Reference Signals
 Artifact Cancellation Using Linearly
Combined Reference Signals

• The vectors can be defined as

and

• The estimate of the EOG artifact is thus obtained by

 Artifact Cancellation Using Linearly
Combined Reference Signals
• All signals are random in nature, with zero-mean, and that

s(n) is uncorrelated with the EOG signals v(n) at each time n,

• the MSE Ew can alternatively be expressed as

• which makes it clear that the weights w should be

chosen such that the error between 𝑣𝑜(𝑛) and 𝑤 𝑇 𝑣(𝑛) is

minimized.
 Artifact Cancellation Using Linearly
Combined Reference Signals

Right eye

Left eye

Before cancellation

After cancellation
 Artifact Cancellation Using Filtered
Reference Signals
• Eye movement and blinks exhibit different spectral properties

when transferred to the EEG.

– Eye movement is transferred at lower frequencies (6 or 7Hz),

– Blinks are transferred at higher frequencies ranging up to the alpha

band (8-13 Hz)

• Improved artifact cancellation can be done by replacing each

of the weights wi by a transfer function hi(n) of a LTI

system-which models the frequency-dependent transmission

of the EOG activity into the EEG

Artifact Cancellation Using Filtered
Reference Signals
 Artifact Cancellation Using Filtered
Reference Signals
• the estimate v0(n) is obtained by filtering each vi(n) with

hi(n), followed by summation of the M filter outputs,

• Each filter is defined by a finite impulse response (FIR)

with length L

• To find the optimal FIR filters hi, the MSE is now defined

by
 Spectral Analysis of EEG signals
• Spectral analysis can be used to determine whether a specific

rhythm is present or not.

• Fourier-based spectral analysis is commonly used

• The power spectrum of the stationary signal x(n) is defined

• Computing the squared magnitude of the N-point DTFT of

x(n) and efficiently implemented by the FFT algorithm.

Spectral Analysis of EEG signals

Spectral analysis of an EEG with alpha rhythm

 Evoked Potentials (EP)
• Evoked potentials (EPs) constitute an event-related activity

which occurs as the electrical response from the brain or the

brainstem to various types of sensory stimulation of

nervous tissues; auditory and visual stimulation are

commonly used.

• Provides information on:

– sensory pathways
– abnormalities,
– the localization of lesions affecting the sensory pathways,
– disorders related to language and speech.
 Evoked Potentials (EP)
• Individual EPs have a very low amplitude, ranging from 0.1

to 10 μV, and are, accordingly, hidden in the ongoing EEG

background activity,

• The repetitive stimulation can be used in combination with

ensemble averaging techniques to help reduce the noise level

• With a sufficiently low noise level, the time delay (latency)

and amplitude of each constituent wave of the EP can be

accurately estimated and interpreted in suitable clinical terms

 Evoked Potentials (EP)

Various morphologies of evoked potentials.

• The duration, amplitude, and morphology differ considerably

from potential to potential.

Ensemble Averaging
 Evoked Potentials (EP)
• For all modalities, measurements on latency and amplitude are extracted

from the waves of the averaged EP and are compared to normative

values in order to discriminate normal, healthy subjects from subjects

with various kinds of neurological impairment

• Factors which suggest that an EP should be interpreted as abnormal

include waves which

– have increased latency,

– have decreased amplitude, or

– are missing.