EEG Analysis MethodsCourtesy Dr Thomas F. Codura

Aamir Saeed MalikNeuro-Signal Processing GroupUniversiti Teknologi PetronasMalaysia

EEG Analysis Methods

Digital Filtering (IIR/FIR) Butterworth, Chebyshev etc Fast response, uses predefined bands (like 7 to

12 hz for alpha) Like using a colored lens Fast, useful for training & assessment Digital clinical norms, live Z-scores based on

digital filters (like joint time frequency analysis)

EEG Analysis Methods

Fast Fourier Transform (FFT) Analyzes all frequencies Like a prism decomposes all lights Response is slower because a block of data is

required for FFT. Typically, its 1 sec past data & looking at mid of that half sec

Not useful for training because digital filtering can give immediate result for biofeedback & we can respond immediately

Its useful for assessment because it can be done a little later. We generally have time to make assessment

EEG Component Bands

EEG Bands

Typical EEG component bands: Delta (1-4 hz) Theta (4-7 hz) Alpha (8-12 hz) Low Beta (12-15 hz) Beta (15-20 hz) High Beta (20-30 hz) Gamma (40 hz and above) Ranges are typical, not definitive Anyone of these bands can occur outside the

above frequency ranges There may be overlap between these bands

EEG Bands

An example: In this room, anyone above 5 ft 10 inches is

male and below it is female If true then we set these frequency ranges If false then I change this figure, say 5 ft 9

inches If true then we set these frequency ranges If false then we change again If false such that one female is 5 ft 10 inches

and one male is 5 ft 9 inches then we throw both of them out

That is, they are outliers and we reject them

EEG Bands

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 In a certain phase of sleep, rapid eye movement called

(REM) sleep, the person dreams and has active movements of the eyes, which can be seen as a characteristic EEG signal

In deep sleep, the EEG has large and slow deflections called delta waves

No cerebral activity can be detected from a patient with complete cerebral death.

Delta Band

It is a very diffuse slow wave It is generally seen in sleep It is often seen in injury, coma, brain trauma Its is not sinusoidal, it is like a wandering wave It is not rhythmic typically It is caused by a variety of mechanisms,

including slow neuronal mechanisms, DC shifts caused by Glial cells

It is often called an injury potential There is a larger delta that is seen in waking

up because eye opening, moving, muscle movement also show up in this EEG

Delta Band

It is a very diffuse slow wave It is often seen after epileptic seizures,

the EEG is seen to wander around Skin potentials can show up as delta Slow cortical potentials can show up as

delta

Theta Band

It is typically from 4-7 hz It is non-sinusoidal wave It is wandering and looks like a square wave It is mediated by the sub-thalamic connections

to the cortex It tends to indicate either distractibility or a

deep inward awareness or lack of presence Its very common in attention deficit disorder &

other types of disorders where brain excitability is not well regulated

It is also associated with memory retrieval It is also associated with extreme creativity

Theta Band

It is typically from 4-7 hz Experienced meditators tend to exhibit theta

during their meditation It indicates an internalized state and a state of

inward focus In epilepsy & attention deficit people, an excess

theta is seen Since it is also associated with artistry & creativity,

we are looking for appropriate amount of theta So we look at ratios like theta to beta etc Also alpha waves can slow down to delta range Alpha slows down with age. The older you get, the

more time you spend processing

Theta Band

It is typically from 4-7 hz So if a person is sitting with eyes closed

& EEG is recorded Lets say its 7 hz waxing & vanning

sinusoidal. What is it? Its alpha because theta is not waxing &

vanning. It just come & go

Alpha Band

It is typically from 8-12 hz Alpha is the resting rhythm of the visual system It is maximal at occipital When eyes are closed, its largest in the back at

occipital It increases when eyes closed It is very sinusoidal, very nice & smooth It indicates that the visual system is at rest &

not processing information There is a reverberation between thalamus &

cortex It indicates that memory scanning is occurring

Alpha Band

It is typically from 8-12 hz Its background memory scanning (not the willful

memory scanning like remembering phone no.) Even there is memory scanning that you are not

aware of. The brain quietly process information sorting it out

It could be slower than 8 hz & faster than 12 hz It also does occur in other locations like frontal

alpha & temporal alpha Alpha are healthy rhythms & they need to be

present at certain level (this will help us using Z-scores & normative data)

You need to have right amount of alpha

Alpha Band

It is typically from 8-12 hz Connectivity measures are important with alpha

like coherence, synchrony and asymmetry By looking at these measure of alpha, they are of

tremendous diagnostic importance For example, alpha synchrony is very healthy in an

individual & it is also trained as biofeedback parameter even in people who are perfectly normal & want to have improved mental performance

Everyone can benefit from alpha synchrony training at the back of the head

James Hart has done it for decades & got good results

Alpha Band

It is typically from 8-12 hz There are variants of high alpha that are abnormal For example, if you see large amounts of diffuse

alpha in an individual & that alpha is seen to drone on & on – indicates anxiety disorder

That brain is adapting to chronic anxiety by saying that I am going to check out, I am not going to deal with it, I am not going to think or process etc. I am just going to make alpha waves

So it’s a paradox because this is a rhythm of relaxation, achieved in meditation while at the same time could indicate tension & anxiety

Excess alpha could also be some compensatory mechanism compensating for some other thing

Alpha Band

It is typically from 8-12 hz So, have to look at global &

comprehensive picture using clinical science

So normative data & experience in reading recordings is very important

Low Beta Band

It is typically from 12-15 hz Some people call it 14 hz like Michael Tansey

who is pioneer in this field He used filter with pass range 13 to 15 hz so

as to center on 14 hz Some people use 12 to 16 hz Low beta can also be called high alpha

because names are not as important as to what's happening

This rhythm when observed over motor strip (C3, C4 & Cz) is called SMR (Sensory Motor Rhythm)

Low Beta Band

It is typically from 12-15 hz SMR is the alpha waves of the motor system It is maximal at C3, C4 & Cz and it is maximum when

body is still Dr Barry Sterman discovered this rhythm in cats who

were sitting still SMR is maximum when brain intends to remain still It is very exciting because it is a measurement that we

can train (biofeedback & teach) for intention (not just behavior but also intention)

SMR deals with what I intend to do, what I plan to do So when it is up-trained, it was found that individual

becomes quiet, still, relaxed & are focused & remain in concentrated state

Low Beta Band

It is typically from 12-15 hz The state where lot of SMR is being produced, it’s a

very studious, quiet, healthy & pensive state You can do maths problems in that state You can read in that state You can sit through a lecture in that state (It means

your legs are not jiggling, you are not shuffling around in chair, you are not wandering about lunch & when the class is over)

Why its 12 to 15 hz? The thalamic-cortical reverberations between motor

strip & thalamus takes about 80 millisec from one way to other & then take 80 ms to get back

Low Beta Band

It is typically from 12-15 hz The distance is that much shorter than the

distance of thalamus to the back of head (Occipital)

The frequencies of these rhythms are directly related to the anatomy & physiology

So the reason that SMR is little bit faster than alpha is quite simply the fact that it is a little shorter trip to go from thalamus to central lobe compared to thalamus to occipital lobe

Beta Band

It is typically from 15-20 hz Presence of beta is typically considered a thinking state As we move to high frequency, the waves become more

& more localized For example, alpha could involve both hemispheres &

you could have synchronous alpha for both hemispheres. People train whole head alpha

Low beta is typically more localized to one hemisphere or other (low beta synchrony across hemispheres is not done. Infact, it is asymmetry between hemispheres)

Beta is even more localized in a certain area The reason is that the connections in brain that produce

the beta waves are more cortical-cortical Its one part of cortex talking to neighboring part in

cortex

Beta Band

It is typically from 15-20 hz So beta waves tend to be localized where the

work is being done Beta activity may be associated with memory

recall when reading (more activity over motor strip)

Again there is a normal amount of beta If you don’t see normative amount of beta, that

means there is under activity (person may not be sharp enough, feeling depressed etc)

Beta is the sign of activation (how active you feel)

You need to use caution when you up-train beta

Beta Band

It is typically from 15-20 hz Why use caution in up-training beta? Generally, you are telling some part of brain to activate

but you are not telling some other part of brain not to activate

Its like poking a tiger with hot rod – you don’t know if he is going to bite or run off or attack

There is too much going on when you try & excite the brain by up-training beta

If done rightly, the results very good in terms of brain sharpening & mood elevation

It should not be done in people prone to anxiety, mood problems etc

An alternative safer way is to down-train a low frequency. As a result, brain naturally shifts to high frequency

High Beta Band

It is typically from 20-30 hz People even go upto 35hz At high beta, muscular activity can sneak in

(EMG) The only way to tell the difference between high

beta & muscle activity is by looking at waves Muscle activity is more like droning activity while

high beta is more like waxing & vanning, true sinusoidal

High beta is associated with intensive thinking, worrying about things like doing a hard maths problem

High Beta Band

It is typically from 20-30 hz Generally, guard (stop) bands used here so that

artifacts like teeth grinding, muscular activity etc does not leak down to alpha

With kids, it is very important to observe them. They might make slight faces when nervous or otherwise and hence get desired waves & get reward

So it is important to tell them in detail & then also to observe them, otherwise, you might get false data

You need to tell them to be quiet, still, relax, listen/see & then ring the bell (for example)

High Beta Band

It is typically from 20-30 hz With theta, its more like drifting away With high beta, its thinking to solving

problems Normative data has been shown to

distribute very well across cultures and regions all over the world

Everybody's brain is built pretty much the same way

Gamma Band

It is typically from 40 hz & above There is more attention put on it since 2000 People have gone uptil 64hz & call it gamma It indicates high level of sensory & perceptual binding It may be an indicator that binding has occurred

(researchers not sure) What is binding? Lets take a pen – I can move it in my fingers (motor

interaction), I can see it from any side, I can hide behind & just feel it – but I know it’s a single object that is pen

There is no confusion in my brain what it is & brain sees it a single entity

Gamma Band

It is typically from 40 hz & above Now little baby child do not have this ability When its gone out of sight, its gone When it comes again, its something new to them We can smell ink of pen & we can hear if it has some

button etc So there is information coming from all of my sensory

mechanisms (sight, sound, smell, touch, taste, memory)

To me this is this pen – That’s Binding That’s sensory/perceptual/Memory Binding On computers, we use these features for object

identification but still we are far from it Naturally, we do it immediately

Gamma Band

It is typically from 40 hz & above A paper on Buddhist meditation shows that the gamma

waves are synchronous across brain This paper from Richard Davidson lab in Univ of

Wisconsin Its miraculous that these high freq waves (40cps) can

actually be synchronized across the brain The brain can’t transfer information that fast There is underlying mechanism (sub-threshold

mechanism) that takes care of this synchrony It’s a theory that there are some sub-threshold gamma

which become synchronized when some event occurs Since it become synchronized all over the brain and all

at once so it has to do with sensory & information binding

Gamma Band

It is typically from 40 hz & above A native American healer was found to

have 100hz gamma There is also relationship reported

between gamma & theta When theta comes up then we see a

burst of gamma

Typical EEG Metrics

EEG Metrics

Typical EEG metrics include: Amplitude (microvolts) Frequency (hz, peak or modal) Percent energy Variability Coherence between 2 channels (percent) Phase between 2 channels (degrees or percent) Asymmetry between 2 channels (ratio or

percent)

Possible Research Topic: New metric/feature for EEG assessment

EEG Metrics

Explanation of EEG metrics: Power (microvolts squared) Amplitude looks at how big it is. We can have

instantaneous amplitude at any time instant or we can have average amplitude

Typically average amplitude is over 1 to 3 sec For assessment, average amplitude is even over a minute There is phase relation between channels (exactly where

the peak & valleys lie – are they in same phase or are their some separations?)

In some cases, phase separation is normal. For example, in language areas, the information goes back & forth when you talk/listen/read

So they do not operate synchronously and there is normal phase difference

EEG Metrics

Explanation of EEG metrics: Asymmetry is very important – it tells us how

different 2 channels are in amplitude Most notable is the frontal alpha where the right

alpha tends to be 10 to 15% larger than left alpha This indicates a difference in cortical activation If the brain waves at the frontal cortex are not

asymmetrical, you will have mood problems (mood disorder, sad person)

Richard Davidson, Peter Rosenfield and few researchers from North Western in Chicago have worked in this and developed protocols

They have used it as therapeutic to assist people who are struggling with mood

EEG Metrics

Explanation of EEG metrics: The underlying rationale is that left hemisphere is

responsible for approach behavior & left frontal lobes are responsible for planning & executing approach behavior

The right frontal lobes are responsible for planning & execution of avoidance behavior

So the lobes for avoidance want to be less activated compared to lobes for approach behavior

You can switch when appropriate but it is important to have bias

Hence the right hemisphere makes more alpha so the right hemisphere is more relaxed

The left hemisphere is more activated The general rule is that to improve mood, activate left

more

EEG Metrics

Explanation of EEG metrics: So it means inhibit theta/alpha/beta on left

hemisphere Push things down a little bit That’s preferential then to enhance things

on right because if you enhance things on right, you are not telling brain not to do it on left

As a result, brain tends to enhance both sides & unexpected results might occur

It has been found that asymmetry stretches to all bands except gamma

EEG Metrics

Explanation of EEG metrics: Same is true for central lobes, C3 & C4 We want C3 to have little more

speed/high frequency than C4 Typically, beta is trained on C3 and SMR

on C4 The brainmaster have built in protocol

that does that – it up-trains 15 to 18% at C3 & 12 to 15% at C4

So left hemisphere speeds up a little bit & right hemisphere relaxes a little bit