Pulse Oximeter Working Principle: How SpO2 Is Measured

Pulse oximetry is one of the most used monitoring methods in medicine. The WHO Pulse Oximetry Training Manual describes it as a standard monitor in anaesthesia and critical care, and small fingertip devices are now common in clinics and homes. For engineers, it is a neat example of an optical sensor, an analog front end and signal processing working together, and it builds directly on the ideas in biopotential measurement basics.

What a pulse oximeter measures

Haemoglobin in red blood cells carries oxygen. Oxygen saturation is the fraction of haemoglobin that is oxygenated:

$$S = \frac{[HbO_2]}{[HbO_2] + [Hb]} \times 100\%$$

where HbO2 is oxygenated haemoglobin and Hb is deoxygenated (reduced) haemoglobin. When this is measured on an arterial blood sample with a laboratory CO-oximeter it is called SaO2. When it is estimated through the skin by a pulse oximeter it is called SpO2. The device also reports pulse rate, taken from the time between pulses in the optical signal.

Pulse oximeter working principle step by step

1. Light absorption by haemoglobin

The method rests on the Beer-Lambert law, which states that the absorbance of light passing through a medium is proportional to the concentration of the absorbing substance and the path length. The same law is used in a spectrophotometer. Webster’s Design of Pulse Oximeters explains that HbO2 and Hb have different absorption spectra:

• At red wavelengths near 660 nm, deoxygenated haemoglobin absorbs much more light than oxygenated haemoglobin.
• At infrared wavelengths near 940 nm, oxygenated haemoglobin absorbs slightly more than deoxygenated haemoglobin.

Because the two forms behave in opposite ways at the two wavelengths, the ratio of red to infrared absorption changes strongly with saturation.

2. Separating arterial blood from everything else

Light passing through a finger is absorbed by skin, bone, tissue, venous blood and arterial blood. Most of these absorb a constant amount. Arterial blood volume, however, rises and falls with every heartbeat as the arteries expand. The detected light therefore has two parts:

• A large, steady DC component from tissue, bone, venous blood and the non pulsatile part of arterial blood.
• A small AC component that pulses at the heart rate. It comes only from the changing volume of arterial blood.

This pulsatile optical signal is the photoplethysmogram (PPG). By using only the AC part, the oximeter isolates arterial blood without knowing the absorption of the tissue.

3. Time multiplexed LEDs and one detector

A typical transmission probe has a red LED and an infrared LED on one side of the finger and a single silicon photodiode on the other. Webster describes how the LEDs are switched on in turn, with a dark period in which both are off. The dark period measures ambient light so it can be subtracted. A transimpedance amplifier turns the photodiode current into a voltage, which is filtered and digitised. Since a photodiode has a broad spectral response, multiplexing in time is what lets one detector measure both wavelengths.

Download diagram (WebP)

4. The ratio of ratios

For each wavelength the processor divides the AC amplitude by the DC level. This normalisation cancels the effect of LED brightness, skin thickness and detector gain. The two normalised values are then divided to give the ratio of ratios R:

$$R = \frac{AC_{red} / DC_{red}}{AC_{IR} / DC_{IR}}$$

As saturation falls, red absorption by Hb rises, so the red AC component grows relative to the infrared one and R increases.

5. Calibration curve

Light in tissue scatters, so the simple Beer-Lambert law does not give SpO2 directly from R. Webster explains that manufacturers build an empirical calibration curve by recording R from healthy volunteers while a CO-oximeter measures SaO2 on arterial samples as the inspired oxygen is lowered in controlled steps. The resulting curve, stored as a lookup table or polynomial, maps R to SpO2. Texas Instruments application report SLAA274 uses a simple linear approximation for a demonstration design:

$$SpO_2 \approx 110 - 25R$$

This is suitable only for learning. Clinical devices use their own validated curves.

Accuracy and limitations

ISO 80601-2-61:2017 is the international safety and performance standard for pulse oximeter equipment. It requires manufacturers to state SpO2 accuracy as a root mean square difference (Arms) against SaO2 from controlled desaturation studies, and it sets a maximum Arms of 4 percent over the range 70 to 100 percent SpO2. Factors that reduce accuracy, described in the WHO manual and the FDA safety communication, include:

• Poor perfusion: cold hands, shock or vasoconstriction make the AC signal very small.
• Motion artefact: movement changes the optical path and can mimic a pulse.
• Dyshaemoglobins: carboxyhaemoglobin and methaemoglobin are not separated by two wavelength devices.
• Skin pigmentation: the FDA noted in 2021 that pulse oximeters may be less accurate in people with darker skin pigmentation.
• Nail polish and ambient light: some dark nail polish absorbs light, and bright room light can saturate the detector.
• Low saturation: calibration data are sparse below 70 percent, so readings there are less reliable.

Worked example: SpO2 from a PPG signal

Notice that the AC components are only about 1 to 2 percent of the DC levels. That is why the analog front end needs low noise and good ambient light rejection, and why the signal is filtered and averaged.

Types of pulse oximeter probes

Design points for engineers

• LED drive: constant current pulses, with adjustable intensity so the DC level sits in the ADC range for thin and thick fingers.
• Front end: a transimpedance amplifier with a large feedback resistor converts nanoamp to microamp currents into volts. Ambient light is subtracted using the dark phase.
• Sampling: the PPG has most of its energy at low frequencies, so moderate sampling rates are enough, but the sampling theorem still sets the minimum rate after filtering.
• Processing: band pass filtering, peak detection, AC and DC estimation, averaging and signal quality checks before a value is displayed.
• Regulation: pulse oximeters for clinical use are regulated medical devices. Their class depends on the market, as explained in medical device classification.

Applications

• Continuous monitoring during anaesthesia, recovery and intensive care.
• Spot checks in wards, clinics, ambulances and triage.
• Home monitoring of people with lung or heart disease on medical advice.
• Sleep studies, where SpO2 dips are recorded overnight.
• Wearable fitness devices using reflectance sensors, which are generally not validated to clinical standards.

Related reading

• Biopotential Measurement Basics
• Spectrophotometer Working Principle
• Medical Imaging Modalities