MASINDE MULIRO UNIVERSITY OF SCIENCE AND TECHNOLOGY (MMUST)

Introduction to Bioelectric Signals

Course: ECE331 – Introduction to Biomedical Engineering
Department of Electrical & Communication Engineering – 3rd Year Undergraduate
Comprehensive Study Guide • Lecture-Aligned • With Worked Examples, Diagrams & Self-Assessment

Learning Objectives

By the end of this unit, the student should be able to:

1. Origin of Bioelectric Signals

1.1 What Are Bioelectric Signals?

Bioelectric signals are electrical potentials generated by living cells, arising from the movement of ions (primarily Na⁺, K⁺, Ca²⁺ and Cl⁻) across cell membranes. These signals are the "electrical language" of the nervous system, the heart, and muscle tissue. For an electrical engineer, they represent extremely low-level, low-frequency analogue signals that must be measured non-invasively through the skin — a challenge that defines the entire field of biomedical instrumentation.

Key Concept: Every bioelectric signal originates from a transmembrane potential difference. The cell membrane acts as a capacitor ( lipid bilayer ≈ 1 μF/cm² ) separating charge, while ion channels act as voltage- and ligand-dependent conductances. The measured surface signal is the summation of thousands of such cellular "batteries" acting in synchrony.

1.2 The Resting Membrane Potential

In its resting state, an excitable cell maintains a potential difference of approximately −70 mV (inside relative to outside). Three mechanisms establish and maintain this potential:

  1. Ion concentration gradients — maintained by the Na⁺/K⁺ ATPase pump, which moves 3 Na⁺ ions out and 2 K⁺ ions in per ATP molecule consumed;
  2. Selective membrane permeability — at rest the membrane is ~50–100× more permeable to K⁺ than to Na⁺ (leak channels);
  3. Impermeant intracellular anions (proteins, phosphates) that trap K⁺ inside the cell.
Nernst Equation (at 37 °C):   Eion = (61 / z) · log₁₀( [ion]outside / [ion]inside )  [mV]

Typical mammalian values: E_K ≈ −90 mV, E_Na ≈ +60 mV, E_Cl ≈ −70 mV. Because the membrane is most permeable to K⁺ at rest, the resting potential (−70 mV) lies close to E_K.

Goldman–Hodgkin–Katz (multi-ion) Equation:   Vm = 61 · log₁₀( (P_K[K⁺]o + P_Na[Na⁺]o + P_Cl[Cl⁻]i) / (P_K[K⁺]i + P_Na[Na⁺]i + P_Cl[Cl⁻]o) )
Worked Example 1.1 — Nernst calculation for K⁺. Given [K⁺]in = 100 mM, [K⁺]out = 5 mM, z = +1: E_K = 61 · log₁₀(5/100) = 61 · (−1.301) ≈ −79 mV. Interpretation: if the membrane were permeable only to K⁺, the potential would stabilise at −79 mV — the K⁺ equilibrium potential.

1.3 The Membrane as an Electrical Circuit

An excitable cell membrane is modelled electrically as a parallel RC circuit: the lipid bilayer is the capacitor Cm; ion channels are conductances g_Na, g_K, g_Cl in series with their Nernst "batteries"; the pump is a current source. This equivalent circuit is the foundation for cable theory and for understanding how signals propagate and how extracellular potentials arise.

2. The Action Potential

2.1 Phases of the Action Potential

When a stimulus depolarises the membrane to threshold (≈ −55 mV), voltage-gated Na⁺ channels open and an action potential (AP) is fired — an "all-or-none" pulse lasting ~1–2 ms in nerves, ~200–300 ms in cardiac myocytes.

PhaseMembrane EventIonic BasisApprox. Potential
RestingSteady stateHigh K⁺ permeability; Na⁺/K⁺ pump active−70 mV
DepolarisationRising phaseRapid Na⁺ influx through voltage-gated Na⁺ channels (positive feedback)−55 → +30 mV
RepolarisationFalling phaseNa⁺ channels inactivate; K⁺ efflux through voltage-gated K⁺ channels+30 → −70 mV
HyperpolarisationUndershootK⁺ channels slow to close≈ −90 mV
Refractory periodsAbsolute / relativeNa⁺ channel inactivation (absolute); K⁺ efflux still high (relative)—
t V (mV) 0 −70 Resting Depolarisation Repolarisation Hyperpolarisation
Figure 2.1 — Typical action potential waveform showing depolarisation, repolarisation and hyperpolarisation phases relative to the −70 mV resting potential.

2.2 Propagation and Extracellular Recording

Because of the local-circuit currents between active and adjacent resting membrane, the AP propagates unidirectionally along an axon at a speed that increases with fibre diameter and myelination (saltatory conduction). Importantly, what surface electrodes measure is the extracellular field potential produced by the flow of ions in the tissue volume conductor — the summation of many simultaneously active cells. This explains why the ECG, EEG and EMG are superpositions of many cellular events, not single action potentials.

Key Concept: The signal-to-noise problem in biopotential measurement begins here — the extracellular summation of ~10⁵–10⁶ cells is needed before a millivolt-level signal appears at the skin. Any loss of synchrony (arrhythmia, muscle fatigue, epilepsy) changes the recorded waveform's amplitude and morphology.

3. Major Bioelectric Signal Modalities

Four bioelectric signals dominate clinical and engineering practice. Table 3.1 summarises their defining characteristics.

SignalSourceTypical AmplitudeFrequency RangePrimary Clinical Use
ECGHeart (cardiac myocytes / conduction system)1 – 5 mV0.05 – 150 Hz (diagnostic)Arrhythmias, ischaemia, conduction defects
EEGBrain (cortical pyramidal neurons)1 – 100 μV0.5 – 100 HzEpilepsy, sleep staging, anaesthesia depth
EMGSkeletal muscle (motor-unit action potentials)0.1 – 5 mV20 – 2000 HzNeuromuscular disorders, prosthetics, ergonomics
EOGEye (cornea–retina dipole)10 – 100 μV0.1 – 10 Hz (DC–10 Hz)Eye tracking, sleep research, vestibular testing

Table 3.1 — Characteristics of the principal bioelectric signals. Always verify band limits against the relevant clinical standard (e.g., IEC/AAMI for ECG).

3.1 Electrocardiogram (ECG)

Amplitude: 1–5 mV | Bandwidth: 0.05–150 Hz | Surface origin: thorax

Each cardiac cycle produces the sequence: P wave (atrial depolarisation) → QRS complex (ventricular depolarisation; atrial repolarisation is hidden within it) → T wave (ventricular repolarisation). The 12-lead system uses limb leads (I, II, III, aVR, aVL, aVF — derived from Einthoven's triangle) and precordial leads V1–V6 to view the cardiac dipole from multiple spatial directions. The ECG's millivolt amplitude makes it the most forgiving bioelectric signal to acquire.

3.2 Electroencephalogram (EEG)

Amplitude: 1–100 μV | Bandwidth: 0.5–100 Hz | Surface origin: scalp (10–20 system)

Rhythmic cortical activity is classified into bands: delta (0.5–4 Hz, deep sleep), theta (4–8 Hz, drowsiness), alpha (8–13 Hz, relaxed eyes-closed, maximal occipitally), beta (13–30 Hz, alert eyes-open), gamma (>30 Hz, cognition). Its microvolt amplitude and susceptibility to 50/60 Hz interference make EEG the most demanding signal to record cleanly.

3.3 Electromyogram (EMG)

Amplitude: 0.1–5 mV | Bandwidth: 20–2000 Hz | Surface or needle electrodes

The surface EMG is the summation of motor-unit action potentials (MUAPs). Force is graded through recruitment (adding motor units) and rate coding (increasing firing frequency), so RMS EMG amplitude correlates with muscle force — the principle behind myoelectric prostheses and fatigue monitoring.

3.4 Electrooculogram (EOG)

Amplitude: 10–100 μV | Bandwidth: DC–10 Hz | Cornea(+)–retina(−) dipole

The eye behaves as an electric dipole; rotating the gaze rotates the dipole, producing a signal roughly proportional to the angle of deviation (≈ 5–20 μV/°). EOG is used in sleep staging (rapid eye movements), vestibular testing and assistive eye-tracking interfaces.

4. Electrodes and the Electrode–Skin Interface

4.1 Surface Electrodes

Surface electrodes transduce ionic current in the body into electron current in the measuring circuit. The dominant type is the Ag/AgCl electrode: silver coated with silver chloride provides a stable, non-polarisable half-cell with a small, reproducible half-cell potential (~+0.223 V vs. standard hydrogen electrode) and low noise.

Key Concept — Polarisation: A perfectly polarisable electrode (e.g., platinum in saline) behaves like a pure capacitor — charge builds up and no charge crosses the interface, causing drift and artefact. A non-polarisable electrode (Ag/AgCl) freely exchanges Cl⁻ ions, so the interface behaves resistively and stable DC/low-frequency measurements become possible.

4.2 Electrode–Skin Impedance and Motion Artefact

PropertyAg/AgCl Gel ElectrodeDry / Textile Electrode
Skin–electrode impedanceLow, stable (~5–10 kΩ)Higher, less stable
Motion artefactLowSignificant
Wear durationLimited by gel drying (hours–days)Long-term (weeks–months)
Patient comfort / prepRequires skin prep; gel may irritateNo prep; most comfortable
Best suited forHigh-fidelity clinical recordingsLong-term / ambulatory / wearable monitoring

Table 4.1 — Engineering trade-off between gel and dry electrodes.

5. Signal Acquisition: The Biopotential Front-End

5.1 The Acquisition Chain

A complete biopotential measurement system can be modelled as a cascade of stages, each of which must be specified for the target signal:

Electrodes (Ag/AgCl) Protection & Lead Selection Instrumentation Amplifier (CMRR) Filtering HP / LP / Notch ADC & Sampling (Anti-alias, Nyquist)
Figure 5.1 — Block diagram of a complete biopotential acquisition chain: electrodes → protection/lead selection → instrumentation amplifier → filtering → ADC/sampling.

5.2 Differential Amplification and CMRR

Because bioelectric signals appear as potential differences between two electrode sites, they are measured with a differential amplifier whose output is proportional to (V⁺ − V⁻). Any interference picked up equally by both leads (mains hum, electromagnetic coupling) is a common-mode signal that the amplifier should reject.

CMRR = 20 · log₁₀( Adifferential / Acommon-mode )  [dB]

Clinical-grade instrumentation amplifiers require CMRR ≥ 100 dB (a factor of 10⁵ in amplitude). CMRR is degraded in practice by electrode impedance mismatch between the two leads — which is why low, balanced electrode impedances matter as much as amplifier design.

Worked Example 5.1 — Effect of CMRR. An amplifier has A_d = 1000 and A_cm = 0.1 (CMRR = 80 dB). A 200 mV common-mode mains artefact produces 0.1 × 200 mV = 20 mV at the output — still four times larger than a 5 mV ECG! Practical systems therefore combine high CMRR with shielding, driven shields, right-leg drive, and notch filtering to reach millivolt-level recording fidelity.

5.3 Noise and Interference Management

Interference SourceFrequency CharacterMitigation Strategy
Mains power (50/60 Hz)Narrowband at line frequency + harmonicsHigh CMRR, shielded cables, notch filter, proper grounding
Electrode half-cell driftVery low frequency (near-DC)AC coupling / high-pass filter at 0.05–0.5 Hz
Motion artefactLow frequency, large amplitudeSecure electrodes, band-pass filtering, signal quality indices
Electromyographic contamination (EEG)20–2000 HzLow-pass below EMG band; patient relaxation
Thermal (Johnson) noiseBroadband whiteLow source impedance; minimal bandwidth
Amplifier input noiseBroadband (nV/√Hz)Select low-noise front-end; adequate gain distribution

Table 5.1 — Common interference sources and standard mitigation strategies.

5.4 Filtering for Each Modality

6. Digitization and the Nyquist Theorem

After amplification and filtering, the analogue signal is converted to digital form by an analogue-to-digital converter (ADC). Two parameters govern digitization: sampling rate and resolution.

Nyquist Sampling Theorem:   fs > 2 · fmax

Undersampling causes aliasing — high-frequency content folds back into the signal band and becomes indistinguishable from genuine low-frequency activity. The anti-aliasing filter is therefore placed before the ADC and is mandatory, not optional.

Worked Example 6.1 — EEG sampling. Diagnostic EEG bandwidth ≈ 100 Hz → minimum sampling rate = 200 Hz. Clinical systems use 250–512 Hz for margin. Sampling at 2000 Hz would add no fidelity to a 10 Hz alpha rhythm, but would increase storage and processing by 8× — an example of engineering judgment over specification excess.

6.1 ADC Resolution

An N-bit ADC divides the full-scale range into 2N levels, giving a quantization step of VFS/2N. For a ±5 V range with a 16-bit ADC, the step is ≈ 0.15 mV — insufficient for EEG unless preceded by adequate gain. Resolution must therefore be specified together with front-end gain (effective resolution referred to the input).

6.2 Typical Acquisition Specifications

SignalFront-End GainBand-PassTypical Sampling Rate
ECG×100 – ×5000.05 – 150 Hz250 – 1000 Hz
EEG×10,000 – ×50,0000.5 – 100 Hz250 – 1000 Hz
EMG (surface)×100 – ×100020 – 500 Hz1000 – 2000 Hz
EOG×500 – ×50000.1 – 10 Hz50 – 200 Hz

Table 6.1 — Reference acquisition specifications used in practice (adapt to equipment and standards in the laboratory).

7. Applications and Clinical Relevance

Engineering Perspective: The same front-end architecture (differential amplification, high CMRR, anti-alias filtering) recurs in every modality — only the gain, band-pass and electrode arrangement change. Mastery of this one signal chain is the transferable skill of biomedical instrumentation.

8. Summary of Key Points

  1. Bioelectric signals arise from ionic currents across cell membranes; the resting potential (−70 mV) results from K⁺ permeability and the Na⁺/K⁺ pump, quantified by the Nernst and Goldman equations.
  2. The action potential is an all-or-none depolarisation–repolarisation cycle driven by voltage-gated Na⁺ then K⁺ channels; surface recordings are the extracellular summation of many synchronised cells.
  3. The four principal modalities differ by two orders of magnitude in amplitude (ECG/EMG in mV; EEG/EOG in μV) and in bandwidth (EOG near-DC to 10 Hz; EMG to 2 kHz) — these numbers drive the entire instrument design.
  4. Ag/AgCl non-polarisable electrodes minimise drift and noise; electrode–skin impedance and motion artefact are the dominant practical problems.
  5. Differential amplifiers with CMRR ≥ 100 dB reject common-mode mains interference; electrode impedance imbalance is the practical limit on CMRR.
  6. Filtering is modality-specific: high-pass removes electrode drift and baseline wander; low-pass defines diagnostic bandwidth; notch filters target line frequency.
  7. The Nyquist theorem (f_s > 2f_max) mandates anti-alias filtering before the ADC; sampling-rate and resolution choices must be justified by signal bandwidth and amplitude, not by specification excess.
  8. A single acquisition-chain architecture generalises across ECG, EEG, EMG and EOG — gain, filtering and electrodes are the variables.

9. Self-Assessment Questions

1. Explain why the resting membrane potential is closer to E_K than to E_Na.
At rest, the membrane permeability to K⁺ is ~50–100 times greater than to Na⁺; the Goldman equation weights each ion's contribution by its permeability, so V_m (−70 mV) lies near E_K (−90 mV).
2. A differential amplifier has A_d = 10,000 and A_cm = 1. Compute the CMRR and explain its meaning in the context of a 100 mV mains artefact superimposed on a 20 μV EEG signal.
CMRR = 20 log₁₀(10⁴/1) = 80 dB. The artefact appears at the output as 100 mV / 10⁴ = 10 μV — half the EEG amplitude. An 80 dB amplifier alone is insufficient; additional mitigation (notch, shielding, right-leg drive) is required.
3. Why is AC coupling (a high-pass filter) mandatory for ECG recording, and what diagnostic information could a very aggressive high-pass filter destroy?
AC coupling removes the electrode half-cell potential mismatch and baseline wander (hundreds of mV) that would otherwise saturate the amplifier. Setting the high-pass corner too high (e.g., above 0.5–1 Hz) distorts the ST segment and can mask ischaemic changes.
4. An EMG system samples at 250 Hz. What happens to a 300 Hz motor-unit component, and what specification change is required?
300 Hz exceeds the Nyquist limit (125 Hz), so it aliases to |250 − 300| = 50 Hz, appearing as false low-frequency content. The sampling rate must be raised above 600 Hz (with an anti-alias low-pass filter at ~250 Hz), or the bandwidth limited before sampling.
5. Compare gel and dry electrodes for a 24-hour ambulatory ECG monitor, and justify a final design choice.
Gel electrodes give lower impedance and less motion artefact but dry out and irritate skin; dry electrodes tolerate long wear but give higher impedance and more artefact. For 24-hour ambulatory use, either long-wear hydrogel Ag/AgCl or dry textile electrodes with active buffering are defensible; the choice should be justified against the monitor's noise floor and artefact-tolerance algorithms.
6. Why does a 12-lead ECG provide diagnostic information that a single-lead smartwatch ECG cannot?
Twelve leads sample the cardiac dipole from multiple spatial directions (frontal plane via limb leads + transverse plane via precordials), enabling localisation of ischaemia and conduction block. A single lead views the dipole from one direction only — adequate for rhythm classification (e.g., AFib) but not for spatial diagnosis.

10. References and Further Reading

  1. Webster, J. G. (ed.), Medical Instrumentation: Application and Design, 4th ed., Wiley — Chapters on biopotential electrodes and amplifiers.
  2. Bronzino, J. D. & Peterson, D. R. (eds.), The Biomedical Engineering Handbook, 4th ed., CRC Press.
  3. Kandel, E. R. et al., Principles of Neural Science, 5th ed., McGraw-Hill — membrane potentials and synaptic transmission.
  4. IEC 60601-1 and IEC 60601-2-47 standards — medical electrical equipment and ambulatory ECG requirements.
  5. Bergey, G. E., "Biopotential Amplifiers," in The Biomedical Engineering Handbook, CRC Press.
  6. Semmlow, J. L., Circuits, Signals, and Systems for Bioengineers, 3rd ed., Academic Press.