As communities and businesses explore ways to protect against community spread of COVID-19 as they open back up, demand for thermal imaging technology for screening individuals for high temperatures is in high demand.
FierceElectronics spoke with Dimitrios Damianos, PhD., a Technology and Market Analyst with the Photonics, Sensing & Display division at Yole Développement (Yole) about the promise for the technology and what the future holds.
FE: In the fight against COVID, why is MEMS technology gaining attention?
Damianos: In the fight against COVID-19, there are various technologies gaining attention, each one driven by a specific application or need. One example of such application/need is the early detection of possible COVID-19 symptoms, amongst which is a fever. The purpose is to isolate febrile people for safety reasons and then provide further diagnostic testing to verify if they have the virus or not. For that reason, thermal imaging and sensing has gained a lot of attention. It allows for fast, contactless screening of body temperature in various non-medical environments. The thermal imaging/sensing application uses MEMS technologies: notably for thermal imaging, microbolometers are used and for simple temperature sensing/detection (no imaging function), thermopiles are used. Both of these devices are based on MEMS technologies.
FE: Where is MEMS applied in thermal detectors and thermal imaging technology?
Damianos: Actually, MEMS devices are used in order to build the end-systems: thermal cameras and temperature guns/contactless thermometers. Thermal cameras use microbolometer focal plane arrays (FPA) with more than 10,000 pixels, up to 1 million pixels (or even more), where each individual pixel can read the temperature thereby creating an image showing temperature variation (a thermal image). Temperature guns and forehead/contactless/IR thermometers use thermopile detectors that measure only the temperature from a specific target/spot, ideally the forehead. However, they do not possess imaging functions since they only have from 1 pixel up to a few (or few thousands) of pixels.
FE: Is thermal camera technology basically “ready to go” for use in screening for elevated body temperature, or does it require some additional enhancements?
Damianos: Some thermal cameras are ready to go, but others are not. The ones that are ready to use for screening elevated body temperatures are the ones already designated for this purpose as medical grade devices and which have acquired FDA-510k (or similar) approval. At the moment, only FLIR (the global thermal camera leader) and a couple of other firms (Infrared Cameras INC, Optotherm) offer these cameras, which have been developed for more than 15 years, with the outbreaks of previous epidemics (namely SARS in 2003, H1N1 in 2009, Ebola in 2014, Zika in 2015). Their initial use was in airports and similar structures, mainly to detect people with fever before boarding a plane. But the size of these previous epidemics was not as big as the current one, for this application (fever detection) and technology (microbolometer) to make headlines.
Other thermal cameras with no FDA approval, typically used in thermography/surveillance applications, can potentially be used for detecting elevated body temperatures. The waters are a bit blurry right now because the FDA has communicated that it will not oppose thermal camera sellers/manufacturers that do not have 510K approval. From one point of view this is justified, in order to let users have easy access to these systems since demand is huge and supply is scarce. Nevertheless, the FDA suggests that the use of these cameras, when used for elevated body temperature detection, must comply with specific specifications. Some examples (non-exhaustive list):
- People must be scanned one by one. The camera works better if it focuses only on one person. In a crowded place people are at various distances and may therefore be out of focus which gives less reliable measurements.
- Temperature must be taken from the inner canthus (eye tear duct) since it’s the most representative spot of the inner/body temperature.
- The environmental conditions must be taken into account, as they can alter the measurement. For example, in winter the surface temperature of the skin might read lower, therefore people must acclimatize for some time before being measured. The same happens during the summer, only the temperatures outside are higher.
- Temperature accuracy must be well below 0.5°C, ideally 0.1-0.2°C, to avoid false negatives. To ensure that, blackbody or similar calibration of the camera should be used.
- Resolution must be high enough to capture miniscule details, such as the face and the focus on isolated regions of the face, such as the tear ducts as mentioned previously.
- The cameras must do radiometric measurements (measure absolute temperature), as non-radiometric cameras also exist (measure relative temperature ) which do not have the required accuracy.
FE: What about the gold rush? Is there a lot of junk out there?
Damianos: There is a real gold rush around thermal cameras. It is only natural that companies try to profit from this, by proposing solutions that they think are good. This is done mostly due to naivety and lack of experience with thermal imaging, because many of these companies, as integrators/resellers, will buy mainly cheaper components in order to make a profit. And these components do not always have the right properties. But there must also be education of the end-users, to teach them to be careful when they choose a system for their own use-case.
At Yole Développement (Yole) we think it would start to gain traction in airports to provide some sense of safety to passengers (even if it’s a minor one). This will allow isolation of flagged people for further testing. Then it would spread to other structures, such as businesses, schools, train/bus stations, etc., where a lot of people gather together at any one time. We have already seen the use of thermal cameras in such cases.
FE: Is the technology going to need AI to take a step further in people identification (and let’s not even go into the privacy issues here!)
Damianos: The application itself (thermal imaging for fever screening), doesn’t need AI for people identification. Besides the privacy issues, there is no need to identify the person since the main use of these cameras is for detection/prevention and access control, and for that you need to simply tell reliably if a person has a fever or not. There is no need to associate the temperature with the person who has that temperature. For the moment, companies offer AI solutions but on bi-spectrum cameras (combination of CMOS sensor for normal images & microbolometer for thermal images). And their use is to identify the person and associate the temperature with their identity, or to focus on the inner eye area as well as isolate the face from other objects around the target and in the background. But the AI works on the visible image, not the thermal one.
However, there are some baby steps from Intel, which is trying to identify people using thermal images only. In order for AI to work, whatever the application, very high-resolution cameras are needed, along with a powerful processing chip and optimized software (ideally neural networks).
FE: Can you talk about a couple of real-world use cases today?
Damianos: Thermal imaging, while not widely known to the masses, is used in many different applications today. It originated in traditional military applications, such as night vision, enhanced driver vision (in armored vehicles), missile seekers, etc., but there are also numerous commercial applications: automotive night vision (now going to ADAS), industrial thermography, perimeter surveillance, security, border control, maritime patrol, firefighting, gas leaks/fire detection, personal vision systems (for example scopes for hunters).
FE: Bottom line, can we expect a lot more temperature screening as economies start to re-open?
Damianos: We would say so, but it will depend on geography as well. We are really seeing a surge in the demand for these systems, initially originating in China and elsewhere in Asia and then spreading to the US and finally Europe. Privacy concerns in EU/US are high, which will slightly delay the deployment of such thermal imaging systems. For sure, they can be used as a first line of defense against COVID-19 (along with other measures), allowing businesses and economies to start again. Thermal imaging could become even more relevant especially in a situation where new daily cases surge or a 2nd wave appears, and new lock-down measures are put in place where an economy continues to suffer.
At Yole Développement (Yole) we have a bold outlook: thermal imaging will come to the masses. First it will penetrate the smartphone segment starting from simple temperature detection/sensing function (no imaging capabilities, justified by the cheap price of such sensors). Then, as consumers become accustomed to this functionality and also more educated, new use-cases could be found and economies of scale could allow a thermal imager to reach a really low price and be integrated in smartphones. It will be something very similar to what happened with CMOS image sensors.
Dimitrios Damianos is delivering a keynote on Thursday, July 16, at 11 a.m. ET as part of Sensor Innovation Week, a free virtual event. Visit the event website for more details and to register for this event.