01Executive summary
Between 1899 and 1910 three developments made the living body an object of measurement, of controlled energy delivery and of remote manipulation. Biomedical engineering begins here.
In 1899 Jean-Louis Prévost and Frédéric Batelli demonstrated at Geneva that a controlled electric shock could terminate ventricular fibrillation. In 1903 Willem Einthoven recorded the heart’s electrical activity with a string galvanometer and established the labelling of the waveform features still used today; he received the Nobel Prize for the work. In 1910 Hans Christian Jacobaeus performed endoscopic examination of the thorax, separating the surgeon’s eye and hand from the operating site.
02Measuring a millivolt through skin
The engineering difficulty of the electrocardiograph is entirely a signal integrity problem. The wanted signal is of the order of one millivolt at the body surface. It is buried in electrode half-cell potentials that can be hundreds of times larger, in muscle activity, in movement artefact and — today — in mains interference coupled capacitively into the patient and the leads.
Einthoven had no usable amplifier, so he built a transducer sensitive enough not to need one: a fine silvered quartz filament suspended in a strong magnetic field. Current through the filament deflects it, and the shadow of the filament is projected and recorded. Patients immersed their limbs in saline vessels to obtain a low-impedance electrical connection. The apparatus was heavier than the patient and needed several operators.
What engineers did next, and why it is the interesting part
The science was finished in 1903. Everything since has been engineering, and it is a textbook sequence of the discipline doing what it does:
Replace the transducer with amplification
A differential amplifier with very high input impedance and high common-mode rejection extracts a small difference between two leads while rejecting the much larger signal common to both. This is why the interference does not swamp the trace.
Eliminate the saline vessels
Silver / silver-chloride gel electrodes provide a stable, low-noise, low-drift interface with minimal half-cell potential. Electrode chemistry is a large part of why modern ECG works.
Filter deliberately
High-pass filtering removes baseline wander from respiration; notch filtering attenuates mains interference; low-pass filtering limits muscle noise. Every filter distorts the waveform, so diagnostic bandwidth is standardised rather than left to the designer.
Digitise and interpret
Sampling and digital processing enable automated rhythm classification, storage, transmission and — critically for the defibrillator — a machine decision about whether a shock is indicated.
Any device electrically connected to a patient must limit leakage current under both normal and single-fault conditions, because the connection may bypass skin resistance entirely. Electrical isolation of the patient circuit, defined creepage and clearance distances, and applied-part classification are design requirements from the first schematic, not certification paperwork added at the end.
03Defibrillation: controlled energy delivery
In ventricular fibrillation the heart’s coordinated electrical activation degenerates into disorganised local activity. Mechanical output effectively ceases. The therapeutic aim is counter-intuitive: not to restart the heart but to stop it completely by depolarising the myocardium simultaneously, so that the natural pacemaker can resume coordinated rhythm.
- AcquireAdhesive electrodes pick up the surface ECG; impedance across the pads is measured.
- AnalyseAn algorithm classifies the rhythm and decides whether a shock is indicated — a machine diagnosis.
- ChargeA converter charges a high-voltage capacitor from a low-voltage battery to a specified stored energy.
- DeliverA controlled biphasic waveform is discharged through the chest, with duration compensated for measured impedance.
Why the automated external defibrillator matters more than the shock
Survival from out-of-hospital cardiac arrest falls sharply with every minute of delay. No clinical arrangement can put a trained clinician beside every patient within minutes. The engineering answer was to move the expertise into the device: automate the diagnosis, reduce the mass from tens of kilograms to a few, make the interface usable by an untrained bystander under extreme stress, and drive unit cost low enough that devices can be distributed by the tens of thousands into public places.
Notice what kind of problem that is. It is not a medical problem or an electrical one. It is industrial design, human factors, embedded software, power electronics, reliability engineering and cost engineering, applied together to a clinical objective. That combination is the definition of the biomedical engineering discipline.
Medical devices supplied in Australia must be included in the Australian Register of Therapeutic
Goods, administered by the Therapeutic Goods Administration, with conformity assessment scaled to
device classification. Electrical medical equipment is designed and tested to the
AS/NZS 3200 and IEC 60601 families, which set requirements for leakage
current, isolation, applied-part type and essential performance. For software that performs a
clinical function — such as shock advisory logic — the software itself may be regulated
as a medical device, with development lifecycle requirements aligned to IEC 62304.
Treat regulatory classification as an early design input; it determines architecture, not just
documentation.
04Keyhole surgery: designing out the incision
Traditional open surgery accepts a large incision because the surgeon needs direct vision and direct manual access. Every consequence of that incision — infection risk, pain, muscle division, scarring, extended recovery — is a cost paid for access, not for the procedure itself.
Jacobaeus’s 1910 thoracoscopy reframed the requirement. If vision can be provided by an instrument and manipulation by long tools, the access opening only needs to admit the instruments. Several openings of a centimetre or less replace one of many centimetres.
| Lost capability | Engineering substitute | Residual difficulty |
|---|---|---|
| Direct vision | Rod-lens or chip-on-tip camera with fibre-optic illumination | Two-dimensional view; loss of depth perception |
| Working space | Controlled gas insufflation lifts the abdominal wall clear of the viscera | Pressure must be regulated and monitored continuously |
| Hand dexterity | Long instruments through fixed ports — grasp, cut, seal, suture | Fulcrum effect reverses motion; degrees of freedom reduced |
| Tactile feedback | Largely absent; substituted by visual cues and experience | The hardest remaining gap; a major driver of robotic systems |
| Haemostasis | Electrosurgical and ultrasonic sealing devices | Thermal spread to adjacent tissue must be controlled |
The breakthrough was not a better instrument. It was recognising that the incision was serving the surgeon, not the patient — and that a requirement inherited for practical reasons can be removed if the underlying need is met another way. Asking “what is this requirement actually for, and who benefits from it?” is one of the most productive questions available in any design review, and it is exactly the question that separates genuine innovation from incremental improvement.
05Takeaways
Small signals need rejection, not gain
Common-mode rejection and electrode quality matter more than amplification. Fix the noise before boosting the signal.
Move expertise into the device
Automating the diagnosis is what let defibrillators be deployed where clinicians are not.
Interrogate inherited requirements
The large incision existed to serve access, not treatment. Trace each requirement to the need behind it.
Regulation is an architecture input
Classification and isolation requirements shape the design from the first sketch. Introduce them early.