Engineers Can Develop Fever-Sensing Technology. But Should We?
Thermal imaging technology has become ubiquitous in industrial and commercial applications. Now, governments and health regulators are recommending temperature checks as a method to combat the spread of COVID-19.
Professional engineers and certified technologists are bound by a code of ethics that legally binds them to uphold the highest standards of design safety as it pertains to the public. But sometimes, the tenets of public safety (an engineer's duty of care) may seem to clash with the code of ethics to which we are bound.
In May 2020, the Center for Disease Control (CDC) issued a series of recommendations pertaining to "return-to-work" programs across the country. The most controversial of these recommendations asks employers to consider performing daily health checks on their employees, including temperature checks. As designers, we can develop the technology to apply modern thermal imaging to expedite this in the workplace—but should we?
FLIR technology in use at the Incheon Airport in South Korea. Image used courtesy of FLIR
Putting the specific ethical COVID-19 concerns aside, we can question the simple efficacy of using this technology to provide health checks of employees. Recently, Jonah M. Kessel, a video journalist for The New York Times decided to take a stroll through Maplewood, N.J. in infrared. Kessel is dubious about using IR cameras to combat COVID-19, and suggests that migrating this technology from an industrial process to monitoring public health is rife with hurdles—technological and ethical.
Basic Teardown of a Thermal Imager
According to FLIR, a thermal imaging company, thermal imagers consist of four major blocks: 1) optics to focus the infrared energy, 2) a thermal sensor to receive the focused energy into an analog front-end (AFE) receiver, 3) a DSP processor to condition signals, and 4) some type of display output.
For the circuit designer, the focus is on improving the AFE. Design specifications such as filtering, amplification and linearity, and analog-to-digital conversion are central to the design of an AFE. Texas Instruments demonstrates a reference design for the AFE of a thermal imaging camera. The design features a high SNR, and sufficient bit density for high-resolution conversion within the ADC. TI also has reference designs for camera power supply and bias (as seen below).
Bias supply generation diagram for ultra-low noise power supply reference design of a thermal image sensor. Image used courtesy of Texas Instruments
Challenges of Wide-Scan Temperate Checks
The design challenges for implementing wide-scale thermal imaging in a workplace are multitudinous.
Instead, let’s focus on the principal physics issue for all radiated signals: wave propagation. The key parameters affecting propagated radiation include 1) signal strength (related to distance and absorption), 2) signal-to-noise ratios (signal coherency), 3) line-of-sight challenges (multipath issues), and 4) DSP signal conditioning (engineering) at the receiver.
Infrared radiation, which we colloquially associate with body heat, is subject to all of the limitations associated with all wave propagation.
Chris Bainter, Director of Global Business Development for FLIR states, “The problem with crowd scanning is we know temperature measurements are impacted by the distance from camera to target, and crowds are different distances away.”
Additionally, infrared radiation is heavily absorbed by water and CO2 molecules in the troposphere, eroding signal strength and further limiting the effectiveness of sensor technology at range to make precise medical measurements. SNR and LoS are the two remaining critical issues when attempting to isolate a single individual in a sea of workers. Each worker is "noise" to the reference worker, and each worker physically blocks the reference worker.
This leaves engineers with a difficult task: using algorithms and fancy DSP techniques to treat each worker as a discrete "signal" and effectively separate them from the larger dataset.
A Solution for Accuracy: Retina Scanning
Today, mass monitoring seems unlikely to be effective, but there exists a possible technical solution that Kessel mentions in his article. His proposed solution—retina scanning—could significantly improve accuracy. Kessel used his FLIR camera to monitor his wife sitting on his porch and sitting at a table (as seen below), and there was a four-degree difference between the two tests.
Thermal measurement efficacy as a function of distance, demonstrating that closer is better when taking an individual temperature. Image (modified) used courtesy of FLIR and Johan M. Kessel
Although this is not empirical, it clearly illustrates the challenges that Chris Bainter alluded to with distance measurement.
Implementing retina scanning at building entrances could provide for a host of functionality, including temperature sensing. The device could discreetly let the employee know to go home, maintaining privacy. It could also replace traditional access cards. The time to market for such a technology would be minimal since the required underlying subsystems already exist.
When you consider the four mentioned design challenges, all impacting the efficacy and reliability of infrared measurement, wide-scan thermal imaging is unlikely to be effective in combating pandemics like COVID-19. As engineers and technologists, we must strive to maintain our mandate to protect the public, but we must do so in the most ethical way possible.
Have you worked on any devices that are purpose-built to combat COVID-19 or help workers return to work? Share your experience in the comments below.