Pulse Oximeters
Pulse oximeters are medical devices that non-invasively measure blood oxygen saturation (SpO2) and pulse rate using optical sensing technology. These compact devices have become essential tools in clinical settings and increasingly in home health monitoring, providing immediate feedback about respiratory function and cardiovascular status without requiring blood samples.
The electronics in pulse oximeters exploit the different light absorption of oxygenated and deoxygenated hemoglobin to calculate oxygen saturation. Understanding the optical principles, signal processing techniques, and hardware design shows how these devices reach clinically useful accuracy despite significant technical challenges, and where that accuracy ends.
Principles of Pulse Oximetry
Pulse oximetry exploits the different optical absorption spectra of oxyhemoglobin and deoxyhemoglobin. Oxygenated hemoglobin appears bright red because it absorbs less red light and more infrared light, while deoxygenated hemoglobin appears darker and absorbs more red light relative to infrared. By measuring light transmission at both wavelengths, the proportion of oxygenated hemoglobin in the sum of oxygenated and deoxygenated hemoglobin can be calculated.
The measurement uses red light at approximately 660 nanometers and infrared light at approximately 940 nanometers. At 660 nm, deoxyhemoglobin absorbs significantly more light than oxyhemoglobin, while at 940 nm, the relationship reverses. The two absorption curves cross near 800 nm, an isosbestic point at which absorption does not depend on saturation, so one wavelength is chosen on each side of it, where the two hemoglobins differ in opposite directions. The ratio of absorption at these two wavelengths provides the basis for oxygen saturation calculation.
Critical to pulse oximetry is isolating the arterial blood signal from the larger static absorption of tissue, venous blood, and other components. The pulsatile nature of arterial blood flow causes rhythmic changes in light absorption synchronized with the heartbeat; this optical pulse waveform is the photoplethysmogram, and its acquisition is termed photoplethysmography (PPG). By analyzing only this pulsatile component, the measurement reflects arterial oxygen saturation rather than the mixture of arterial and venous blood in the tissue.
The Beer-Lambert law describes how an absorber attenuates light:
A = εcl
where A is the absorbance, ε the extinction coefficient of the absorber at the wavelength used, c its concentration, and l the optical path length. When each heartbeat lengthens the path through arterial blood by a small amount Δl, the fractional change in detected light approximates the resulting change in absorbance. Comparing that change at the two wavelengths cancels Δl, and with it the LED brightness and detector sensitivity, leaving a quantity that depends mainly on the proportions of the two hemoglobins.
The ratio of pulsatile to static components at each wavelength, known as the modulation ratio, therefore varies with oxygen saturation. Tissue scatters light strongly, however, so red and infrared photons travel different effective path lengths, and the idealized model predicts saturation poorly. Empirically derived calibration curves instead relate the ratio of red to infrared modulation ratios to actual oxygen saturation values measured by laboratory co-oximetry. These calibration curves are programmed into each pulse oximeter to convert measured ratios to SpO2 percentages.
A conventional two-wavelength oximeter reports functional saturation, the oxygenated share of the hemoglobin able to carry oxygen, rather than fractional saturation, which counts every hemoglobin species present:
Sfunc = HbO2 / (HbO2 + Hb)
Sfrac = HbO2 / (HbO2 + Hb + COHb + MetHb)
Here each term is the concentration of oxyhemoglobin (HbO2), deoxyhemoglobin (Hb), carboxyhemoglobin (COHb), or methemoglobin (MetHb). The two agree when the dysfunctional hemoglobins are negligible, and diverge in carbon monoxide poisoning and methemoglobinemia, the cases discussed under Limitations and Considerations below.
Optical System Design
Pulse oximeter optical systems consist of light-emitting diodes (LEDs) that illuminate the tissue and photodetectors that measure transmitted or reflected light. The choice of transmission versus reflection geometry depends on the measurement site and device form factor.
Transmission pulse oximeters, commonly used in fingertip devices, place LEDs on one side of a tissue bed such as the finger or earlobe and the photodetector on the opposite side. Light passes through the tissue, and the detector measures how much light of each wavelength makes it through. Because the light crosses the whole vascular bed, this geometry yields a comparatively large pulsatile signal.
Reflection pulse oximeters place LEDs and photodetectors on the same side of the tissue, measuring light that is absorbed, scattered, and reflected back from the tissue. This geometry enables measurement at sites where transmission is impractical, such as the forehead or wrist. Reflection measurements are more challenging due to weaker signals and greater susceptibility to surface reflections and placement variations.
LED selection requires wavelengths closely matched to the optimal measurement points on the hemoglobin absorption spectra. The red LED matters most, because the deoxyhemoglobin absorption curve is steep near 660 nm and a shift of a few nanometers changes the ratio the calibration expects. Manufacturers therefore either select LEDs within tight wavelength tolerances or characterize each LED and apply wavelength-specific calibration corrections; some clinical sensors carry a coding resistor or memory device that tells the monitor which calibration to use.
The photodetector, typically a silicon photodiode, must respond to both red and infrared wavelengths with adequate sensitivity. Silicon is more responsive at 940 nm than at 660 nm, and tissue attenuates the two colors differently, so the drive current of each LED is set separately to bring both channels into a similar part of the measurement range. Ambient light rejection is achieved through opaque sensor housings, optical filtering, and the signal processing techniques described below that separate the LED signals from background illumination.
Signal Acquisition Electronics
The electronic signal chain in a pulse oximeter must detect a small pulsatile change superimposed on a much larger static signal. In a well-perfused finger the pulsation is typically only a few percent of the static level, and under poor perfusion it can fall to a small fraction of one percent. This dynamic range requirement drives the design of the amplification, filtering, and digitization stages.
LEDs are driven in a time-multiplexed sequence, alternating between red, infrared, and off states. The off period allows measurement of ambient light, which is subtracted from the LED-on measurements to remove background interference. This technique, called ambient light cancellation, significantly improves signal-to-noise ratio. Each LED is on for only a small fraction of the cycle, which also keeps drive power low.
The photodetector output passes through a transimpedance amplifier that converts the photocurrent Iph to a voltage set by its feedback resistor Rf:
Vout = −IphRf
A photocurrent of 1 µA through a 500 kΩ feedback resistor, for example, produces an output of 0.5 V in magnitude. The amplifier must remain linear from bright ambient conditions to weak pulsatile signals while adding little noise. Automatic gain control may adjust the LED current or the amplifier gain based on signal strength to make good use of the ADC range.
Synchronous detection, where the signal is sampled in synchronization with the LED switching, extracts the signal component corresponding to each LED wavelength while rejecting interference at other frequencies. This technique is implemented either through analog sample-and-hold circuits or through digital sampling synchronized to the LED timing.
Analog-to-digital converters with 16-bit or greater resolution capture the signal for digital processing. High resolution is essential because the pulsatile component represents only a small fraction of the total signal: if the static level fills half the range of a 16-bit converter, or 32,768 codes, a 1 percent pulsation spans only about 330 codes, and a 0.1 percent pulsation about 33. Oversampling and averaging can improve effective resolution beyond the ADC's native capability. Integrated optical front-end chips now combine the LED drivers, transimpedance amplifier, ambient cancellation, and converter in one package for compact and wearable designs.
Digital Signal Processing
Digital processing extracts the pulsatile components from the digitized signals and calculates oxygen saturation and pulse rate. This processing must handle the challenging conditions of real-world measurements, including patient motion, low perfusion, and ambient light variations.
Separation of the pulsatile (AC) and static (DC) components uses filtering techniques. Low-pass filtering extracts the DC component representing total light transmission, while bandpass filtering isolates the pulsatile component in the 0.5 to 5 Hz range, which spans heart rates from 30 to 300 beats per minute. The AC/DC ratio at each wavelength provides normalization that compensates for variations in LED intensity and tissue optical properties.
Pulse detection algorithms identify individual cardiac cycles in the waveform, enabling beat-to-beat SpO2 calculation and pulse rate determination. Detecting pulses reliably despite noise and artifacts is critical for continuous monitoring accuracy. Algorithms may use threshold crossing, peak detection, or template matching approaches.
The ratio of ratios combines the two normalized pulsatile signals:
R = (ACred/DCred) / (ACIR/DCIR)
R falls as saturation rises, because the red pulsation shrinks relative to the infrared one as deoxyhemoglobin disappears from the arterial blood. The empirical calibration curve maps R to SpO2; on typical curves an R of about 1 corresponds to a saturation of roughly 85 percent. The calculation usually spans several cardiac cycles to provide stable readings, with averaging windows chosen to balance responsiveness against noise reduction. A long window steadies the display but can smooth away brief desaturations, which matters in overnight studies.
Motion artifact rejection is essential for accurate measurement during patient movement. Movement displaces venous blood and tissue as well as arterial blood, and the resulting artifact can be as large as the arterial pulse itself, so a ratio computed naively during motion is unreliable. Algorithms test the correlation between the red and infrared waveforms, reject beats whose shape departs from a learned template, and in more advanced designs use adaptive filters that estimate the noise component and subtract it. Accelerometer data may supplement optical signals for artifact detection in some devices.
Calibration and Accuracy
Pulse oximeter calibration curves are developed through controlled studies comparing device readings with reference measurements from laboratory co-oximeters analyzing arterial blood samples. Healthy subjects breathe oxygen-nitrogen mixtures that lower their saturation in steps, with simultaneous pulse oximeter readings and arterial blood samples establishing the relationship.
Calibration studies typically span saturations from 70 to 100 percent, the clinically relevant range for monitoring. Desaturating volunteers further adds risk, so few studies do so; reference data below 70 percent are scarce, the calibration there is less certain, and many devices do not specify accuracy below this level. Some clinical devices include extended calibration for low saturation monitoring in specific applications.
Accuracy is expressed as the accuracy root mean square, Arms, the measure ISO 80601-2-61 uses, computed over the n paired readings of a validation study:
Arms = √[Σ(SpO2,i − SaO2,i)2 / n]
Here SpO2,i is each oximeter reading and SaO2,i the co-oximeter reference drawn at the same moment. Because Arms combines the mean bias with the scatter about it, it is not a tolerance: only about two-thirds of readings can be expected to fall within ±Arms of the reference. If the errors are roughly normal, a reading of 92 percent from a device with an Arms of 3 percent is consistent, at about 95 percent confidence, with an arterial saturation anywhere from roughly 86 to 98 percent.
The 2011 and 2017 editions of ISO 80601-2-61 required an Arms of 4 percent or less over the 70 to 100 percent range. The United States Food and Drug Administration's premarket guidance of March 2013 recommends tighter values: 3.0 percent for transmittance, wrap, and clip sensors, and 3.5 percent for ear-clip and reflectance sensors. Manufacturers of clinical sensors commonly label values of about 2 to 3 percent under no-motion conditions, with wider figures during motion or low perfusion. In the United States, many inexpensive fingertip oximeters, by contrast, are sold over the counter as general wellness products for sports or aviation use, without regulatory review of their accuracy.
Factors affecting accuracy include dysfunctional hemoglobins such as carboxyhemoglobin and methemoglobin that absorb light differently than normal hemoglobin, potentially causing significant measurement errors. Patient factors including skin pigmentation, nail polish, poor peripheral perfusion, and motion can also degrade accuracy.
Regulatory standards, principally ISO 80601-2-61, specify testing methods and performance requirements for pulse oximeter equipment. Under its 2011 and 2017 editions, the controlled-desaturation validation protocol called for at least 200 paired measurements collected from at least ten healthy subjects, with reference saturations spread across the 70 to 100 percent range; the FDA's 2013 guidance adds that at least two subjects, or 15 percent of the pool if that is larger, should be darkly pigmented. In January 2025 the FDA issued draft guidance proposing markedly larger and more diverse study populations, on the order of 150 participants with substantial representation of darker skin tones, and recommending both the Monk Skin Tone scale and the spectrophotometric individual typology angle to characterize pigmentation objectively. A revised edition of the standard, ISO 80601-2-61:2026, followed; drafts of the revision shared publicly had proposed an Arms limit of 3 percent, at least 24 participants with at least a quarter in each of three skin-tone groups, and limits on the difference in bias between skin-tone groups. Devices marketed for medical use must demonstrate compliance through clinical testing and regulatory submission, though requirements and enforcement vary by jurisdiction and device class.
Fingertip Pulse Oximeters
Fingertip pulse oximeters are the most common consumer pulse oximetry devices, featuring an integrated clip design that positions LEDs and photodetectors appropriately when placed on the finger. These compact devices provide spot-check measurements and typically run on small batteries for extended shelf life.
The transmission optical geometry in fingertip devices provides strong signals due to the finger's favorable optical properties and adequate tissue thickness. The fingertip has good perfusion under normal conditions, enabling reliable pulsatile detection in most users.
Display systems in fingertip oximeters show SpO2 percentage, pulse rate, and often a plethysmograph waveform or signal strength indicator. LED, LCD, or OLED displays with large digits enable reading from various angles. Some devices include bar graphs showing relative signal strength or perfusion index.
Power management is simplified in spot-check devices that operate for brief measurement periods. Auto-off functions preserve battery life by turning off the device when removed from the finger. Low battery indicators warn users when replacement is needed to ensure accurate measurements.
Accuracy limitations in fingertip devices can result from cold fingers causing vasoconstriction and reduced pulsatile signal, movement during measurement, and nail polish or artificial nails that attenuate red light more than infrared. User instructions typically advise warming cold hands and removing nail polish before measurement.
Wearable and Continuous Monitoring
Wearable pulse oximeters enable continuous monitoring of oxygen saturation, valuable for patients with respiratory conditions, sleep apnea assessment, and high-altitude activities. These devices must address the challenges of extended wear, motion during daily activities, and power consumption for continuous operation.
Wrist-worn oximeters use reflection geometry, with LEDs and photodiodes on the back of the case measuring the skin beneath it, usually on the top of the wrist. This location is less favorable than the fingertip because perfusion is lower, the pulsatile signal from the capillary bed is weaker, and hand movements produce more motion artifact. Advanced signal processing is essential for acceptable accuracy.
Finger-worn ring devices position sensors around a finger for transmission or reflection measurements while allowing normal hand use. These devices face challenges from the smaller measurement site and motion during activities, but can provide good accuracy when properly designed and positioned.
Overnight oximetry for sleep apnea screening typically uses fingertip sensors with data logging capability. The device records SpO2 throughout the night, and software analyzes the record for the desaturation events characteristic of apneic episodes, commonly summarized as an oxygen desaturation index: the number of falls in saturation of at least 3 or 4 percent per hour of recording. This application requires secure sensor attachment and comfortable wear during sleep.
Integration of SpO2 monitoring into smartwatches and fitness trackers has expanded consumer access to oxygen saturation data. These multi-sensor devices face significant accuracy challenges due to the non-optimal wrist measurement site and intermittent sampling during daily activities, and their SpO2 features are typically positioned as wellness rather than medical measurements.
Perfusion Index and Signal Quality
Perfusion index (PI) indicates the strength of the pulsatile signal relative to the static signal, providing a measure of peripheral blood flow at the measurement site. Low PI indicates weak pulsatile signal that may compromise SpO2 accuracy, while high PI suggests good signal conditions.
PI is the ratio of the pulsatile to the non-pulsatile component, usually taken from the infrared channel and expressed as a percentage:
PI = (ACIR/DCIR) × 100 percent
Monitors typically display values from about 0.02 percent for a very weak pulse to 20 percent for a very strong one. Values above about 1 percent generally indicate adequate signal for reliable SpO2 measurement, while much lower values mark conditions in which accuracy may be degraded. PI varies widely between people and between measurement sites, so a trend in one patient is more informative than a single value.
Signal quality indicators help users and clinicians assess measurement reliability. Displays may show waveform strength bars, quality scores, or warning indicators when signal conditions are poor. These indicators guide users to reposition sensors or address factors affecting signal quality.
Pleth variability index (PVI), a parameter introduced by Masimo, measures respiratory-induced variation in the perfusion index over one or more respiratory cycles:
PVI = (PImax − PImin) / PImax × 100 percent
Large respiratory swings in mechanically ventilated patients can indicate fluid responsiveness. Because it depends on waveform variation over multiple respiratory cycles, PVI is available primarily in clinical monitoring devices.
Limitations and Considerations
Carbon monoxide poisoning presents a dangerous pulse oximetry limitation because carboxyhemoglobin absorbs red light similarly to oxyhemoglobin, causing falsely normal SpO2 readings despite severely impaired oxygen delivery. Standard two-wavelength pulse oximeters cannot detect carbon monoxide; detecting it requires CO-oximetry of a blood sample or a multi-wavelength pulse CO-oximeter.
Methemoglobinemia, whether from genetic conditions or toxic exposures, similarly confounds standard pulse oximetry. Methemoglobin absorbs strongly and about equally at the red and infrared wavelengths, driving R toward 1, which the calibration maps to roughly 85 percent. As methemoglobin rises, SpO2 readings therefore trend toward 85 percent regardless of actual arterial oxygen saturation, overstating a low saturation and understating a high one.
Skin pigmentation can affect pulse oximetry accuracy. Studies have shown that devices tend to overestimate SpO2 in patients with darker skin pigmentation, particularly at lower saturation levels, increasing the risk of occult hypoxemia in which a normal-appearing reading masks dangerously low arterial saturation. A 2020 analysis in the New England Journal of Medicine found occult hypoxemia nearly three times as often in Black patients as in White patients. This finding has prompted regulators to strengthen the skin-tone diversity expected in validation studies. Manufacturers have worked to address the disparity through improved calibration, but clinicians should remain aware of this limitation and confirm borderline readings with arterial blood gas analysis when hypoxemia is suspected.
Anemia has little direct effect on SpO2 accuracy unless it is severe, when readings at low saturation become less reliable. The larger hazard is interpretive: oximetry measures the percentage of hemoglobin that is oxygenated, not the total oxygen content of blood, so a severely anemic patient may have critically low oxygen delivery despite a normal SpO2 if the reduced hemoglobin present is fully saturated.
Intravenous dyes such as methylene blue, which absorbs strongly near 660 nm, cause transiently low readings. Venous pulsation, from a sensor applied too tightly or from conditions such as tricuspid regurgitation, adds venous blood to the pulsatile signal and also lowers the reading.
External light interference from bright sunlight, surgical lights, or flickering artificial light can affect readings despite ambient light rejection techniques. Shielding the sensor from direct light exposure improves accuracy in challenging lighting environments.
Clinical and Home Use Applications
In clinical settings, pulse oximetry is standard for monitoring during anesthesia, surgery, and post-operative recovery. Continuous SpO2 monitoring provides early warning of respiratory compromise, enabling timely intervention. Alarm systems alert clinicians when saturation drops below configurable thresholds.
Hospital ward monitoring uses pulse oximetry for patients at risk of respiratory deterioration, including those with pneumonia, COPD exacerbations, or recovering from sedation. Spot-check measurements supplement vital sign assessments, while continuous monitoring may be warranted for higher-acuity patients.
Home monitoring applications have expanded significantly, driven by consumer device availability and heightened respiratory health awareness. Patients with chronic respiratory conditions use oximetry to monitor their baseline and detect exacerbations. Exercise desaturation monitoring can identify exertional hypoxemia that occurs only during activity.
The COVID-19 pandemic dramatically increased home pulse oximetry use for monitoring patients with respiratory infections. Early detection of hypoxemia enabled timely healthcare contact and hospitalization when needed, though inappropriate use and interpretation also caused concern about unnecessary anxiety and healthcare utilization.
Aviation and high-altitude applications use pulse oximetry to monitor for hypoxemia at altitude. Pilots, mountaineers, and others operating at elevation can use oximetry to assess their oxygenation status and detect early signs of altitude illness requiring descent or supplemental oxygen.
Future Developments
Multi-wavelength oximetry using additional LED wavelengths can measure dyshemoglobins such as carboxyhemoglobin and methemoglobin that confound standard two-wavelength measurements. Some clinical devices already offer this capability, with potential for expansion to consumer applications.
Tissue oxygen saturation monitoring measures oxygenation in deeper tissues rather than only arterial blood, potentially providing earlier warning of tissue hypoxia in critical illness. Near-infrared spectroscopy techniques enable this measurement, with miniaturization ongoing for wearable applications.
Motion artifact algorithms continue to improve, with machine learning now joining adaptive filtering, and better motion tolerance would extend reliable continuous monitoring to more active scenarios. The push for validation across skin tones is likewise prompting manufacturers to reassess sensor and algorithm design.
Integration with telehealth platforms allows healthcare providers to review pulse oximetry data remotely, supporting management of respiratory conditions without office visits. Automated analysis and alerting help identify the patients who need intervention within the large volume of continuous monitoring data.
Conclusion
A pulse oximeter turns two small optical pulsations into a clinically meaningful number through careful LED timing, low-noise transimpedance amplification, high-resolution conversion, and an empirical calibration that maps the ratio of ratios to saturation. Its readings are statistical estimates with known limits: dyshemoglobins, poor perfusion, motion, and skin pigmentation can each move them away from the true arterial value, and validation requirements have tightened in recent years to address those errors.