Optical sensors shine a spotlight into our well-being, illuminating actionable insights that can guide our behavior. "Optical sensing is important because it's extremely versatile. We shine some light into a region that we want to analyze, the light will interact with the analyte, and we look at how that light changes," explained Maxim’s Ian Chen. "This is very non-intrusive and non-destructive."
Chen, an executive director in the company's Industrial & Healthcare Business Unit, covered how optical techniques are being used in bioanalytical applications during his talk at the recent Sensors Expo & Conference in San Jose, California. In optical sensing, he explained, we're measuring how the optical path has changed, examining factors like light intensity, the presence of fluorescence, interference patterns, and different behaviors of light. Conventional household smoke detectors are an example of a commonly encountered sensing system. Measuring particulates obscuring light in an optical chamber, these detectors assume that anything blocking the light is smoke. But they are prone to false alarm. An intelligent optical-sensing smoke detector, on the other hand, would leverage two colors of light, as light scatters differently by color. Using optical sensing provides a faster response as smoke is caught in open air and, since the detector can be sealed, this approach is also quite rugged.
In healthcare applications, optical heart-rate monitoring involves shining a light, usually a green LED, into human tissue. Then the light coming through or back is examined to determine what’s happening within that tissue. When the heart is pulsating, the amount of blood going into the arteries expands and contracts, changing the amount of light absorbed or reflected. When light is received, a signal results that is proportional and synchronous to the human heart beat. The challenge, however, is the fact that skin is not homogenous; its makeup varies depending on each person. Skin consists of multiple layers of tissue, each with their own reflective and transmissive indices. So, optical design software integrates CAD models of tissue with the optical heart-rate monitor data collected to reveal insights. “Sensing is about how we distill signals we want from all the other stuff affecting the signal received. So it’s not just about sensing the signal. We need to understand the biology,” Chen said.
Optical heart-rate monitoring involves shining an LED light source into human tissue to collect insights.
Differences in optical properties of the skin will influence the magnitude and quality of the photoplethysmogram (PPG) signal detected. A PPG is an optically obtained plethysmogram, which provides a volumetric measurement that is commonly used in wearables to monitor vital signs. Designers must also consider factors such as how movement of the wearable will affect the received signal, how air gaps might reduce the intensity of the received signal, how glass covering the signal might affect it, and so on, Chen noted.
There is a lot more information in pulses than heartbeats. If you look at the shape of the signal, every feature could actually be telling us additional information. Using light to interrogate a blood capillary is a 1D representation of a multidimensional phenomenon. With machine learning, we can look at the shape of the pulse, the height of the peak, the latency of the peak—all of this information can be applied to learn other things about the patients," he explained to a full audience.
While the makeup of the skin is, obviously, outside of the designers control, what can be controlled via the system design is the profusion index (PI). The PI is the ratio between the AC portion of the PPG signal and the DC portion of it. Chen explained that the mechanical design of the device, like a wearable, can be done in a way to maximize the PI. For example, multiple LEDs or multiple photodetectors could render a wearable less vulnerable to the effects of incidental movement. Instead of a watch format, a patch or an in-ear design could also be less affected by movement. Applying machine learning to heart rate or PPG signals presents one way to learn more about, and adjust for, the noise in the system.
Using more than one optical sensor can be beneficial as well. When it comes to evaluating optical sensors, Chen discussed some considerations that the ICs should address. On the transmit path, LED driver noise and linearity are among the design concerns, as any noise in the LED power signal would affect the LED signal strength. On the receive path, ambient light cancellation, a wide signal range, and high signal-to-noise ratio are important. Maxim’s ambient light cancellation is notable, he said, because it applies a two-step approach:
- Analog coarse cancellation, where the ambient light level is captured when the LED is off and then subtracted from the photodetector output before sampling for the PPG signal. To avoid saturating the converter, a coarse DC signal is removed before the sampling.
- Digital fine cancellation, where the LED is off and residual DC, AC, and 1/F noise are removed when sampling
Optical sensing continues to advance, Chen noted, with improvements in power consumption in every generation of product. Lower power consumption, in turn, supports capabilities like sensor fusion. "When we look at sensor fusion, there are a couple of ideas to bring up. Wearable sensors can be used to monitor, let's say heartbeat or blood pressure, or they can be used to provide a continuous set of information for the user," Chen said, noting that either use case is valid but comes with very different power consumption profiles.
Wrapping up his talk, Chen noted: "The best technology is actually invisible. When we get up in the morning and we look at the weather report, the weather is actually an extremely complicated phenomenon that is collected by a large number of sensors. However, we've gotten to the point where if someone tells me today will be 78 degrees with no precipitation, it becomes a set of information that tells me how to dress today.
"We haven't gotten there with any wearable wellness device yet. We don't yet have an ecosystem where a wearable wellness sensor is collaborating with your Alexa, your health application, your running app. We are different silos today. There's a lot of work that is still ahead of us that would allow us to think of health as a piece of actionable information like the weather report. That would be the ultimate advancement—to use wearable sensors to improve human life."
Chen closed his talk with a couple of questions for the audience: what innovation do you want us to work on together? How can we take the next step together?
After all, with more enriched information about human wellness and health, we'll all be better off.
Chen has authored two application notes that provide more in-depth detail on optical sensing:
- Using Reflectometry for a PPG Waveform
- How Common Noise and Error Sources Affect Optical Biosensing