Pulse Oximetry Signal Processing

Reference: Chapter 3 of Pulse Oximetry 2nd Edition by John TB Moyle

1. Overview of Pulse Oximeters

The Basics

Pulse oximeters measure oxygen levels in blood by looking at how light is absorbed by the skin and underlying tissues. These devices usually shine light through a finger, toe, or earlobe, though they can also reflect light off the skin for specific applications like during childbirth. Here, we'll focus on the most common type, which transmits light through a body part.

The chosen body part should be relatively transparent to let light pass through. The light wavelengths used are typically between 600 nm and 1300 nm. The goal is to pick wavelengths where oxygenated and deoxygenated forms of hemoglobin (the protein in red blood cells that carries oxygen) absorb light differently. Since we have two forms of hemoglobin to measure, at least two wavelengths are used.

Design Considerations

When designing these devices, the chosen light wavelengths should be easy to generate and cheap. The device also needs a sensitive detector that doesn't require dangerously high energy levels. Some computing power is also necessary to filter out other factors that could affect light absorption and to calculate the actual oxygen level in the blood.

Originally, lasers were considered for generating the light, as they are monochromatic (single-wavelength). However, lasers would make the device expensive and potentially unsafe. Plus, it would require fragile optical fibers to guide the laser light to the skin.

Modern Improvements

The game-changer was the realization that light-emitting diodes (LEDs) could do the job adequately. LEDs are not strictly single-wavelength, but they're good enough for this application. They're also safer and cheaper than lasers. LEDs can be turned on and off quickly, eliminating the need for special filters and allowing a single sensor to measure light at multiple wavelengths. This makes the device more affordable and reliable.

2. The Probe Design in Pulse Oximeters

2.1 Understanding LEDs in Pulse Oximetry

Light-emitting diodes (LEDs) are specialized semiconductor devices. These are designed to emit light of particular wavelengths when electric current flows through them. The trick lies in "doping" the semiconductor material with specific impurities like aluminum, gallium, or indium, which allows engineers to control the light emission.

Initially, pulse oximeters used 660 nm (red) and 940 nm (near-infrared) wavelengths. These choices were practical for manufacturing and effective for measuring different forms of hemoglobin. As technology advanced, a broader range of wavelengths became available, offering more options for device designs.

2.2 Why These Wavelengths?

The 940 nm wavelength was picked because its absorption properties are relatively constant, so small changes in wavelength won't mess up the calibration. On the other hand, 660 nm was chosen because it provides a large difference in the absorption rates of oxygenated and deoxygenated hemoglobin, making it sensitive to changes in oxygen levels.

2.3 Advantages of LEDs

LEDs are favored for several reasons: - Almost monochromatic - Energy-efficient - Low heat emission - Quick response time - Stable wavelength - Intensity is adjustable - Cheaper and safer compared to alternatives like semiconductor lasers

2.4 Probe Mechanics

In a typical pulse oximeter, LEDs are positioned so that their light passes perpendicularly through the chosen body part, such as a finger, and lands on a photodetector. The probe must be snug but not tight, and shielded from outside light. Incorrect placement can result in calibration errors, especially in small children.

probe_components

2.5 Electrical Design Considerations

Recently, there's been an emphasis on reducing electrical and magnetic noise in the probe assembly. Differential amplifiers are now commonly used to cancel out noise, which is essential in medical applications dealing with small signal voltages.

2.6 LED Variability and Calibration

One challenge is the natural variability in LED emission wavelengths. This can be managed in a few ways: 1. Select LEDs that fall within an acceptable range of error. 2. Measure each LED's wavelength and adjust the device's calibration accordingly. 3. Simply ignore the issue, though this can compromise accuracy, especially at lower oxygen saturation levels.

To make probes interchangeable, some manufacturers include a preset calibration component, usually a fixed resistor, that automatically adjusts the device's calibration based on the specific probe's characteristics.

2.7 Photodetector

Positioning and Function

In a pulse oximeter, a single photodetector is responsible for picking up light from both LEDs. Typically placed opposite to the LEDs, it's designed to be sensitive to a specific range of wavelengths. The setup ensures that any extraneous light doesn't interfere with the readings.

Semiconductor Photodetectors

The most commonly used photodetector is a silicon photodiode. While semiconductors usually aim for stability against outside influences, photodetectors deliberately exploit sensitivity to changes in light levels. They have limited bandwidth, which means you have to carefully choose the LEDs and photodetectors to make sure their ranges overlap.

Dynamic Range

Silicon photodiodes offer a large dynamic range, allowing them to detect varying light levels across a broad spectrum. Phototransistors are another option; they're more sensitive but come with higher electrical noise, making them less ideal for some applications.

Sensitivity Variability

Note that the sensitivity of photodetectors can vary depending on the wavelength. This must be accounted for in both the electronic design and the device calibration.

Electrical Considerations

The signal from the photodetector to the pulse oximeter might be extremely low, requiring special care in cable design. The cables are usually shielded with a conductive braid to protect against electromagnetic interference. Some cables may also include a temperature sensor to monitor the probe and skin temperature.

reducing_interference

The above figure illustrates two approaches for amplifying plethysmograph signals in medical applications, aiming to mitigate the impact of electrical and magnetic noise

  • (a) displays a basic method that uses a single signal conductor and a 'common' pathway, often the shielding braid of the probe's cable. This setup is susceptible to electromagnetic interference (EMI), which can overwhelm the small signal voltages.
  • (b) introduces the use of a differential amplifier. This setup employs two identical conductors connected to an amplifier that outputs the voltage difference between them. The advantage is that EMI tends to affect both conductors equally, effectively canceling out the interference. This technique is crucial for small-signal medical applications like ECG, EEG, EMG, and direct BP measurements.

Cable Flexibility

The cable must be flexible and lightweight to allow for natural movements without affecting the probe's performance or causing any mechanical errors.

3. Electronics in Pulse Oximeters

3.1 Overview

The electronic circuits in a pulse oximeter handle multiple functions:

  • Boosting the weak signal from the photodetector
  • Separating the red and infrared signals from the plethysmograph
  • Controlling the LED current flow
  • Equalizing the gain of the two signals
  • Isolating the "arterial" part of the signal
  • Converting both red and infrared signals from analog to digital
  • Calculating the ratio between red and infrared signals
  • Removing unwanted noise or artifacts
  • Calculating and displaying oxygen saturation (SpO2), heart rate, and plethysmogram
  • Managing alarms
  • Recording SpO2 trends

The received signal from the photodetector is weak and gets amplified first. LEDs alternate to emit light, with brief pauses to measure any interfering ambient light. This amplified signal is then split into red, infrared, and an ambient light component. Electronic filters clean up high-frequency noise, making the signals appear continuous.

electronic_components

Ambient light levels, captured during LED 'off' periods, are subtracted to minimize errors. There are four components in the absorption signal, as shown in the figure below:

components_of_absorption_signal

One signal is adjusted so that the red and infrared signals have equal direct current (DC) components. Then, the ratio of their alternating current (AC) amplitudes is calculated. The figure below shows the red and infrared signals before and after DC equalization.

DC_equalization

Earlier pulse oximeters calculated this ratio only once per cardiac cycle, but modern ones do it multiple times for increased accuracy. While you might think applying the Beer–Lambert law would give precise SpO2 values, it's not that simple due to varying conditions like scattering and path length. Many devices use a lookup table to correlate the red-to-infrared ratio with oxygen saturation. The figure below shows the relationship between SpO2 and the red-to-infrared ratio. The lookup table is made by sampling this curve:

SpO2_vs_led_ratio

Algorithms also consider light intensity and minor variations in the path length. Because patients differ in factors like finger size or skin color, LED intensity is adjusted to prevent signal saturation and potential tissue damage. The microprocessor controls this LED current adjustment and introduces a correction factor for small changes in peak wavelength.

3.2 Elimination of Artifacts in Pulse Oximetry

If pulse oximeters displayed raw oxygen saturation (SpO2) data, the numbers would constantly fluctuate and include erroneous readings due to artifacts. Therefore, software in these devices employs statistical averaging techniques to stabilize the output.

The first priority is to remove artifacts arising from physical movements. For instance, Nellcor employs a specific algorithm in its latest oximeters to address this. The algorithm, called OXISMART, is shown in the figure below:

Nellcor_OXISMART

Other manufacturers have their own methods. Ohmeda uses a weighted averaging system that considers three factors: the size of the real-time signal, when the data was captured in the cardiac cycle, and how closely the new reading matches the currently displayed average. If a new reading deviates greatly from the average, it's given less weight in the calculations. This system averages weighted values over the last 3 seconds, smoothing out minor spikes but still allowing for quick adjustments in genuinely fluctuating conditions.

Masimo Corporation employs its patented Signal Extraction Technology (SET). The software charts a time series of detected plethysmographic data, plotting possible SpO2 values against their likelihood based on past data. When there's less noise due to mechanical artifacts, the displayed SpO2 value is more likely to be accurate. During periods of noise, the system uses historical data to select the most probable accurate value.

Advances in sensor technology have also helped reduce errors from both mechanical movements and electromagnetic interference (EMI).

3.3 Status Messages in Pulse Oximeters

In addition to showing oxygen saturation, heart rate, and plethysmograph data, most pulse oximeters display status messages such as "no probe connected," "probe off patient," "noise signal," "searching for pulse," "battery low," and "insufficient signal."

3.4 Alarm Features

Pulse oximeters are equipped with alarms for low oxygen saturation and abnormal heart rates. While some devices include an alarm for high oxygen saturation, this feature is generally discouraged except in specific neonatal applications. When you turn on the device, the alarms are set to default settings, often with oxygen saturation below 95% and a heart rate between 60 and 100 beats per minute. Users can modify these settings, but they revert to defaults when the device is switched off. Some devices even alter the pitch of the auditory pulse beep to quickly alert users to a drop in oxygen saturation levels.

Many pulse oximeters offer built-in software that automatically tracks and graphs oxygen saturation levels over designated time periods, such as one, two, six, or 12 hours. Some devices also allow for direct recording of this data, along with heart rate, onto a chart recorder or computer. This feature is particularly useful in monitoring conditions like obstructive sleep apnea during sleep studies.

3.6 Power supply

Pulse oximeters need a power supply, and the choice of source depends on whether the device will be used away from mains electricity and for how long. The power needs can be broken down into three categories: light-emitting diodes (LEDs), internal electronics, and display methods.

LEDs in the device require between 10 mA and 200 mA, depending on tissue density. They generally operate on a 1:3 duty ratio, meaning they're active for one-third of the time.

The internal electronics, which include preamplifiers, analog-to-digital converters, and microprocessors, can be designed in two ways. One option uses standard integrated circuits, while the other uses low-current components. Low-current components are more sensitive and expensive but enable faster calculations.

Various display methods are available, each with pros and cons. Cathode ray tube displays are versatile but consume the most power. Seven-segment LED displays are highly visible but require a relatively high current. Liquid crystal displays use very low current but are less legible unless backlit, which increases current consumption.

4. Reflection Pulse Oximetry

Reflection pulse oximetry offers certain benefits over its transmission-based counterpart. Clinically, reflection oximeters are versatile and can be used on various body parts like the forearm, thigh, chest, forehead, and cheeks. This makes them particularly useful for patients with poor blood flow or hypothermia. They're also handy post-cardiac bypass surgery, when peripheral blood flow may be compromised. However, securing the probe can be more challenging in conscious or restless patients, potentially leading to more artifacts in the data.

reflection_pulse_ox

In terms of accuracy, reflection pulse oximetry may sometimes fall short compared to transmission methods. Contributing factors include a lower signal-to-noise ratio, less pronounced heart-related signal changes, and less reliable electronic analysis. Tissue swelling, or edema, can also reduce the method's sensitivity. The technology is still useful for assessing blood supply in skin flaps, transplants, and assessing the viability of injured organs or limbs.

Let's compare the absorption spectra using the transmissive method:

absorption_spectra

And the reflective method:

absorption_vs_reflection_spectra

The two bare resemblance.

5. Pulse Oximetry in Magnetic Resonance Imaging (MRI) Environments

MRI poses unique challenges for electronic monitoring due to its high magnetic fields. Standard pulse oximetry equipment isn't just unreliable in such settings; it can get permanently damaged. Additionally, any metal contact with the patient's skin can cause burns and distort MRI images.

To mitigate these issues, specialized pulse oximeters have been developed for MRI settings. These devices house all electronic components, including LEDs and photodetectors, in a central unit away from the magnetic field. Light is then guided to and from the patient through optical fibers, which are minimally affected by the magnetic field. These MRI-compatible pulse oximeters have been shown to operate both effectively and safely, although they must be positioned at least 3 meters away from the MRI magnet due to the presence of ferromagnetic components in the main unit.