EEG Fundamentals

Maksim Parfyonov, MD FRCPC

November 9, 2026

Objectives

By the end of this session, you should be able to:

  • Explain what scalp EEG actually measures and why it misses a lot
  • Name electrodes in the 10–20 system and read a channel label
  • Apply the polarity convention and localize a discharge on bipolar and referential montages
  • Describe a waveform using the standard vocabulary (frequency, amplitude, location, morphology, reactivity)
  • Know what is normal for age and state in a child’s EEG
  • Recognize the common benign variants and artifacts so you don’t over-read them

Why this matters for you

You will rarely read the EEG yourself. You will:

  • Order the right study (routine vs sleep-deprived vs prolonged vs vEEG)
  • Read the report and decide what is actionable
  • Talk to families about “abnormal” results
  • Look at a tracing at 2 a.m. with a fellow and need to follow along

The most common EEG error is not missing an abnormality, it’s calling a normal finding abnormal. Over-read EEGs lead to epilepsy misdiagnosis and unnecessary ASMs.

A little history

  • Hans Berger (University of Jena) published the first human EEG in 1929
  • Described the alpha rhythm (“Berger rhythm”) and its blocking with eye opening
  • Initially met with skepticism; confirmed by Adrian & Matthews (1934)
  • Clinical EEG for epilepsy took off in the 1930s with Gibbs, Lennox and Davis (3 Hz spike-wave in absence)

Where the signal comes from

What does scalp EEG record?

  • Summated postsynaptic potentials (EPSPs/IPSPs) from cortical pyramidal neurons, not action potentials
    • Action potentials are too brief (~1 ms) and asynchronous to summate
    • Postsynaptic potentials last 10s–100s of ms and overlap in time
  • Pyramidal cells are aligned in parallel, perpendicular to the surface → their fields add up
  • Signal must cross CSF, dura, skull and scalp → heavily attenuated and smeared
    • Scalp EEG: ~10–100 µV; cortical surface: ~10× larger
  • A spike usually needs ≥10 cm² of synchronously active cortex to be seen on the scalp

Consequence: a normal scalp EEG does not mean normal brain. Deep, small, or tangentially oriented sources can be invisible.

The pyramidal neuron as a dipole

  • Excitatory input at the apical dendrite → cations flow in → the extracellular space near the dendrite becomes negative
  • Current returns near the soma → extracellular positive
  • Result: a dipole with a negative and a positive end
  • Scalp sees the end that faces it → most cortical discharges are surface-negative

Dipole model; from F. Borlot teaching deck (2019)

EPSPs, IPSPs and the extracellular field

  • EPSPs summate to threshold → action potential; IPSPs hyperpolarize
  • It is the slow synaptic currents, summed over thousands of cells, that the scalp electrode sees

Niedermeyer’s Electroencephalography, ch. 2

Where do rhythms come from?

  • Thalamocortical loops act as pacemakers
    • Posterior dominant (alpha) rhythm
    • Sleep spindles
    • The spike-wave of absence seizures
  • The rate of synchronous afferent input sets the frequency of the cortical wave
  • Loss of normal thalamic/cortical input (structural lesion, encephalopathy) → slowing

Rhythms are about networks. Slowing means dysfunction; it doesn’t tell you the cause.

Recording the EEG

The 10–20 system

  • Four landmarks: nasion, inion, left & right preauricular points
  • Electrodes placed at 10% and 20% of the measured distances
  • Scales with head size, so it works from neonates to adults
  • 19 scalp electrodes + reference + ground (+ ECG)
  • Neonates: reduced montage (fewer electrodes, same logic)

Reading the electrode names

  • Letter = region: Fp frontopolar, F frontal, C central, T temporal, P parietal, O occipital, A auricular
  • Odd = left, even = right, z = midline (“zero”)
  • Numbers increase away from the midline
  • Old ↔︎ new names (you will see both):
    • T3 / T4 → T7 / T8
    • T5 / T6 → P7 / P8

Libenson, Practical Approach to Electroencephalography

Denser arrays: the 10–10 system

  • Adds electrodes between the 10–20 positions
  • Used for epilepsy surgery evaluations, source localization, and when the standard array is too coarse
  • Inferior temporal chain (F9/F10, T9/T10, P9/P10) helps with basal temporal discharges

The differential amplifier

  • Every channel compares two inputs: Input 1 (G1) minus Input 2 (G2)
  • Signals common to both (e.g., mains noise) are rejected (common-mode rejection)
  • What you see is always a difference. There is no “absolute” voltage on an EEG.

An EEG channel is a voltmeter

Each line on the screen = the voltage of Input 1 relative to Input 2, plotted over time.

Libenson, Practical Approach to Electroencephalography

The polarity convention: “negative is up”

  • If Input 1 is more negative than Input 2 → pen goes up
  • If Input 1 is more positive than Input 2 → pen goes down
  • An upgoing deflection in channel “F3–C3” can mean:
    • F3 is negative, or
    • C3 is positive
  • One channel alone can’t tell you which. You need the neighbors.

Libenson, Practical Approach to Electroencephalography

Display settings change what you see

Setting Typical value What it does Pitfall
Sensitivity 7 µV/mm (adults); often 10–15 µV/mm in children Vertical scale Lower number = bigger waves. Asymmetries can appear or disappear
Time base 30 mm/s (10 s per page) Horizontal scale Compressing time makes waves look sharper and faster
Low-frequency filter (LFF) 1 Hz (0.5 Hz in neonates) Removes slow drift Set too high, it erases real delta
High-frequency filter (HFF) 70 Hz Removes muscle/fast noise Set too low, muscle can look like spikes or beta
Notch 60 Hz (N. America) Removes mains noise Masks a bad electrode that should be fixed

When something looks odd, check the settings first.

Frequency bands

Band Frequency Where you’ll see it normally
Delta (δ) < 4 Hz Infants; deep sleep at any age
Theta (θ) 4 – < 8 Hz Young children awake; drowsiness at any age
Alpha (α) 8 – 13 Hz Posterior dominant rhythm (age ≥ ~3 y), eyes closed
Beta (β) > 13 Hz Frontocentral; ↑ with benzodiazepines, barbiturates

Frequencies are given in Hz (cycles per second). To estimate: count waves in one second.

Frequency bands, illustrated

Fourier: every wave is a sum of sine waves

  • Joseph Fourier: any waveform, however complex, can be described as a sum of simple sine waves
  • Real EEG is a mixture of frequencies riding on each other
  • Your eye does a rough Fourier analysis when you “see” delta under alpha
  • The basis of filters, spectral analysis and quantitative EEG (qEEG)

Libenson, Practical Approach to Electroencephalography

Pulling the mixture apart

1 Hz + 2.4 Hz + 11.7 Hz + 34 Hz = the “messy” wave on the previous slide.

The vocabulary: describing any waveform

When you describe anything on an EEG, cover:

  1. Frequency: Hz / band
  2. Amplitude: µV (low < 20, medium 20–50, high > 50 µV, roughly)
  3. Location / field: focal, regional, lateralized, generalized
  4. Morphology: sinusoidal, sharp, spike (< 70 ms), sharp wave (70–200 ms), polyphasic
  5. Rhythmicity and persistence: rhythmic vs polymorphic; intermittent vs continuous
  6. Symmetry and synchrony between hemispheres
  7. Reactivity: eye opening, alerting, stimulation
  8. State: awake, drowsy, asleep (and which sleep stage)

Montages & localization

Two ways to wire the channels

Bipolar Referential
Each channel Electrode vs adjacent electrode Electrode vs a common reference (ear, Cz, average…)
Shows Voltage gradient between neighbors Something close to absolute amplitude at each electrode
Localize by Phase reversal Highest amplitude
Strengths Sharp localization of focal events; little reference contamination Shows true field shape and amplitude asymmetries
Weaknesses Cancellation of widespread fields; end-of-chain problems Reference contamination (an “active” reference distorts every channel)

Average reference = each electrode vs the average of all (or most) electrodes; good for focal fields, contaminated by large ones.

Always look at a finding in more than one montage.

The “double banana” (longitudinal bipolar)

The default reading montage in most labs:

  • Left temporal chain: Fp1–F7, F7–T7, T7–P7, P7–O1
  • Left parasagittal: Fp1–F3, F3–C3, C3–P3, P3–O1
  • (Midline): Fz–Cz, Cz–Pz
  • Right parasagittal: Fp2–F4 …
  • Right temporal: Fp2–F8 …

Front to back, left over right.

Same discharges, two montages

Referential: look for the biggest deflection. Bipolar: look for where the pens point at each other.

From F. Borlot teaching deck (2019)

Think in fields, not electrodes

  • A discharge creates a voltage field on the scalp, maximal at one point and falling off with distance
  • Every real cerebral event has a plausible field: it spreads to neighboring electrodes
  • An event confined to a single electrode with no field → think artifact (electrode pop)

Libenson, Practical Approach to Electroencephalography

Phase reversal on a bipolar chain

  • C3 is the most negative point (−50 µV)
  • F3–C3: Input 2 more negative → pen down
  • C3–P3: Input 1 more negative → pen up
  • The pens point toward each other at the shared electrode, C3
  • Negative phase reversal localizes the maximum

Libenson, Practical Approach to Electroencephalography

The field matters: amplitude follows the gradient

  • Here the field also involves F3 and P3 (−30 µV)
  • Channels next to the maximum show smaller deflections (20 µV difference)
  • Channels at the edge show larger ones (30 µV difference)
  • Bipolar amplitude reflects the gradient, not the absolute voltage

Libenson, Practical Approach to Electroencephalography, Fig 4-18

When the maximum sits between electrodes

  • F3 and C3 are equally negative (−40 µV)
  • F3–C3 is flat (no difference)
  • The reversal happens across the flat channel
  • Maximum lies between F3 and C3
  • An “isoelectric” phase reversal

Libenson, Practical Approach to Electroencephalography

Three bipolar patterns to know

Instrumental phase reversal: reversal at the shared electrode (F3)

Cancellation (recap of the previous slide): the intervening channel (F3–C3) is flat because both electrodes are equally negative

End of chain: the maximum is at the front of the chain (Fp1 ≈ F3, so Fp1–F3 is flat). Every other channel deflects the same way, so there is no reversal

End-of-chain events (Fp, O, temporal) are why you check a referential montage.

Textbook figures, ch. 8 “Polarity and Field Determinations”; from F. Borlot teaching deck (2019)

Your turn: where is this spike?

Phase reversal in F8–T4 / T4–T6 → maximum at T4 (T8): right mid-temporal spike.

Localization rules, on one slide

Bipolar montage

  • Find the phase reversal; the shared electrode is the maximum
  • Pens pointing toward each other (negative reversal) = surface-negative event (most spikes)
  • Flat channel in the middle of a reversal = maximum between electrodes
  • No reversal in the chain = maximum at the end of the chain

Referential montage

  • Highest amplitude = closest to the source
  • Watch for reference contamination: is the reference itself active?

Always: does the field make physiologic sense?

The normal pediatric EEG

The first two questions

Before calling anything abnormal, ask:

  1. How old is the child? (corrected age if < 2 years and preterm)
  2. What state are they in? Awake, drowsy, or asleep (and which stage)

Almost every pediatric over-read is a normal finding interpreted against the wrong age or the wrong state.

  • Children’s EEGs are higher amplitude and slower than adults’
  • Drowsiness in children can look dramatic (hypersynchrony)
  • Many “sharp” normal variants appear in specific age windows

Posterior dominant rhythm (PDR) by age

Age PDR (eyes closed)
3–4 months Emerges, ~3–4 Hz
12 months ~5–6 Hz
3 years ~8 Hz
8–10 years ~9–10 Hz
Adolescence ~10 Hz (adult range 8.5–11)
  • Should attenuate with eye opening (reactivity)
  • Should be symmetric; right is often slightly higher amplitude (up to ~50% asymmetry tolerated)
  • Rule of thumb: ≥ 8 Hz by age 3. A slow PDR for age is one of the most useful markers of diffuse dysfunction

Normal awake findings in children

Posterior slow waves of youth (PSWY)

  • Delta waves fused into the posterior alpha
  • Peak age 8–14 years, uncommon after 20
  • Block with eye opening together with the PDR
  • Disappear in drowsiness
  • Not focal slowing

Libenson, Practical Approach to Electroencephalography, Fig 11-18

Slow alpha variant

  • Notched, semi-rhythmic theta at half the frequency of the PDR (a subharmonic)
  • Alternates with or replaces the normal alpha
  • Reactive to eye opening, just like the PDR
  • Not “background slowing”

Atlas of Pediatric EEG, Fig 1-7

Drowsiness: hypnagogic / hypnopompic hypersynchrony

  • Paroxysmal bursts of high-voltage, rhythmic 3–5 Hz activity
  • On falling asleep (hypnagogic) or arousal (hypnopompic)
  • Seen from ~6 months to ~11 years
  • Can contain sharpened waves → classic mimic of generalized spike-wave
  • Key: right state, no evolution, monomorphic

Libenson, Practical Approach to Electroencephalography, Fig 11-21

Normal sleep architecture

Vertex waves, sleep spindles, K-complexes, POSTS (positive occipital sharp transients of sleep)

When the sleep features show up

Feature Appears Pediatric note
Sleep spindles (12–14 Hz, central) ~6–8 weeks Asynchronous between hemispheres until ~2 years; this is normal
Vertex waves ~5 months Can be very tall and sharp in children; easily over-read as spikes
K-complexes ~6 months Vertex wave + spindle; often with arousal
POSTS Childhood onward Surface-positive, occipital, “check-mark” shape; normal
Hypnagogic hypersynchrony ~6 mo – ~11 y See previous slide

Absence of expected sleep features (e.g., no spindles after 3–4 months) is more concerning than extra sharpness.

Activation procedures

Atlas of Pediatric EEG, Fig 8-17

Hyperventilation (HV) (3 min minimum)

  • High-amplitude diffuse slowing (“buildup”) is normal in children
  • Watch the child: in this tracing the child stopped blowing the pinwheel and stared off. That’s an absence seizure (3 Hz spike-wave), not buildup
  • Provokes absence seizures in most untreated CAE

Photic stimulation

  • Photic driving (time-locked posterior response) is normal
  • Photoparoxysmal response (generalized spike-wave) is abnormal

Sleep / sleep deprivation

  • ↑ yield of epileptiform discharges, especially in genetic generalized and self-limited focal epilepsies

A word on neonates

The neonatal EEG is a different language, and gets its own session.

  • Interpretation depends on post-menstrual age, week by week
  • Normal discontinuity: tracé discontinu in preterm infants; tracé alternant in term quiet sleep
  • Different normal transients (frontal sharp transients, delta brushes)
  • Different filters (LFF 0.5 Hz), sensitivity, and electrode set

For now: don’t apply older-child rules to a neonatal tracing.

Normal variants: don’t over-read them

How to recognize a benign variant

Benign variants are real cerebral activity with no clinical significance. Clues:

  • Appears in a typical age window and state (often drowsiness)
  • Monomorphic: every burst looks the same
  • Does not evolve in frequency, amplitude or location
  • Doesn’t disrupt the background; often no aftergoing slow wave
  • Often arciform (sharp on one side, rounded on the other)

If you’re not sure whether something is a spike, it probably isn’t. True IEDs usually stand out from the background.

Mu rhythm

  • Arciform 7–11 Hz rhythm over central (C3/C4) regions
  • Awake, eyes open
  • Blocks with movement (or the thought of moving) of the contralateral hand
  • Fragments can look like spikes; compare with the full train
  • Often prominent over a skull defect (breach)

Libenson, Practical Approach to Electroencephalography, Fig 11-6

Breach rhythm

  • Higher voltage, sharper, faster activity over a skull defect (craniotomy, burr hole, fracture)
  • Bone no longer attenuates the signal
  • Sharp morphology can mimic epileptiform activity
  • Know the surgical history before reading

Libenson, Practical Approach to Electroencephalography, Fig 11-12

14 & 6 Hz positive bursts

  • Brief bursts of 14 Hz and/or 6 Hz arciform waves
  • Surface-positive spiky component
  • Posterior temporal, best on ear-referential montage
  • Drowsiness / light sleep
  • Peak in adolescence (13–14 y)
  • Not epileptiform

6 Hz spike-and-wave (“phantom”)

  • Brief (~1 s) bursts of tiny spikes and 6 Hz waves
  • Generalized, low amplitude; the spike may barely be visible
  • Drowsiness, adolescents and adults
  • Benign when: female, occipital, low amplitude, drowsy (FOLD)
  • High amplitude, anterior, awake and male (WHAM) carries more association with seizures

Benign epileptiform transients of sleep (BETS / small sharp spikes)

  • Small (< 50 µV), brief (< 50 ms), mono- or diphasic spikes
  • Light sleep (N1/N2), temporal regions
  • Disappear in deeper sleep
  • No aftergoing slow wave, no disruption of background
  • Mostly adults; uncommon in young children

Rhythmic mid-temporal theta of drowsiness (RMTD)

  • Bursts of 5–7 Hz notched, flat-topped or sharply contoured theta
  • Mid-temporal, uni- or bilateral
  • Drowsiness, adolescents/adults
  • Formerly “psychomotor variant”
  • Unchanging frequency throughout the burst; a seizure would evolve

Wicket spikes (for completeness)

  • Arciform 6–11 Hz trains over the temporal region
  • Drowsiness/light sleep
  • Mostly adults > 30, so rare in pediatrics
  • Single wicket fragments are the classic adult over-read as temporal spikes

Libenson, Practical Approach to Electroencephalography, Fig 11-8

Benign variants: summary

Variant Age State Location Key feature
PSWY 8–14 y Awake, eyes closed Occipital Blocks with eye opening with the PDR
Hypnagogic hypersynchrony 6 mo–11 y Drowsy / arousal Diffuse, frontocentral Paroxysmal but monomorphic
Vertex waves > 5 mo Sleep Vertex Tall and sharp in kids
Mu Older child, adult Awake Central Blocks with contralateral movement
14 & 6 positive bursts Adolescents Drowsy / light sleep Posterior temporal Positive, ear-reference
6 Hz spike-wave Adolescents, adults Drowsy Generalized Tiny spike, FOLD vs WHAM
BETS Adults > children Light sleep Temporal Small, brief, no slow wave
RMTD Adolescents, adults Drowsy Mid-temporal Rhythmic theta, no evolution
Breach Any Any Over skull defect Know the history

Artifacts

Artifacts: anything not from the brain

Physiologic

  • Eye movements and blinks (the eye is a dipole: cornea +, retina −)
  • Muscle (fast, high-frequency)
  • Glossokinetic (tongue, chewing, sucking, pacifiers)
  • ECG / pulse
  • Movement: patting, rocking, crying, being held

Non-physiologic

  • Electrode “pops” (single electrode, no field)
  • 60 Hz mains
  • ICU devices: IV pumps, ventilators, ECMO, high-frequency oscillators

The best artifact detector is the video and the tech’s notes.

Eye movements in a neonate

Use EOG channels to separate eye movement from frontal slowing. In neonates, eye movements are also a sleep-state marker (active sleep).

ICU artifacts can look rhythmic

Infusion pump motor: very high frequency, fixed morphology and repetition rate

IV drip: transient with each drop, in an otherwise inactive record

ECMO: rhythmic, regular, device-locked

Clues: too regular, no physiologic field, unchanged by state, and it stops when the device does.

Putting it together

Reading an EEG report

Section What to look for
Background PDR frequency for age, reactivity, symmetry, continuity
Sleep Were normal sleep features present?
Activation HV/photic; any response or clinical event
Abnormalities Slowing (focal vs generalized), epileptiform discharges, events
Impression Normal vs abnormal, and why
Clinical correlation How the reader ties it to the question you asked

Give the reader a good question. “R/o seizure” gets a worse answer than “8 yo with nocturnal facial twitching. ?SeLECTS”.

What a normal EEG does (and doesn’t) mean

  • A single routine EEG shows epileptiform discharges in only about 30–50% of people with epilepsy
  • Yield ↑ with sleep, sleep deprivation, repeat recordings, and prolonged recording
  • A few percent of healthy children have epileptiform discharges (e.g., centrotemporal spikes) and never have a seizure

Epilepsy is a clinical diagnosis. The EEG supports it, classifies it, and guides treatment. It doesn’t make the diagnosis on its own.

Five habits to leave with

1. Age and state first

Judge everything against what’s normal for this child, in this state.

2. Negative is up, and you need the neighbors

One channel never localizes. Find the field.

3. Look in more than one montage

Bipolar for gradients, referential for amplitude and field shape.

4. Sharp ≠ epileptiform

Know the variants. If in doubt, it’s probably not a spike.

5. Check the video and the settings

Most “weird” findings are artifacts or display choices.

Selected references

  • Libenson MH. Practical Approach to Electroencephalography. Saunders/Elsevier; 2010.
  • Schomer DL, Lopes da Silva FH, eds. Niedermeyer’s Electroencephalography: Basic Principles, Clinical Applications, and Related Fields. 6th–7th ed. Lippincott Williams & Wilkins / Oxford University Press.
  • Tatum WO. Handbook of EEG Interpretation. 3rd ed. Demos; 2021.
  • Acharya JN, Hani AJ, Cheek J, Thirumala P, Tsuchida TN. American Clinical Neurophysiology Society Guideline 2: Guidelines for standard electrode position nomenclature. J Clin Neurophysiol. 2016;33(4):308–311.
  • Acharya JN, Hani AJ, Thirumala PD, Tsuchida TN. ACNS Guideline 3: A proposal for standard montages to be used in clinical EEG. J Clin Neurophysiol. 2016;33(4):312–316.
  • Kuratani J, Pearl PL, Sullivan L, et al. ACNS Guideline 5: Minimum technical standards for pediatric electroencephalography. J Clin Neurophysiol. 2016;33(4):320–323.
  • Kane N, Acharya J, Beniczky S, et al. A revised glossary of terms most commonly used by clinical electroencephalographers. Clin Neurophysiol Pract. 2017;2:170–185.
  • Tao JX, Ray A, Hawes-Ebersole S, Ebersole JS. Intracranial EEG substrates of scalp EEG interictal spikes. Epilepsia. 2005;46(5):669–676.
  • Benbadis SR, Tatum WO. Overinterpretation of EEGs and misdiagnosis of epilepsy. J Clin Neurophysiol. 2003;20(1):42–44.