The new buzz in the thermal imaging world, which has exploded into the lives of anyone dealing with thermal imaging, goes by many names. In a short time, a small niche in the world of IR, which was previously sidelined to make way for more lucrative markets such as security and defense, has taken the top spot in the attention, production, and sales for many manufacturers and integrators.
It’s no surprise considering the size of this new market. Suddenly, hotels, cinemas, malls, hospitals, critical services, public transportation, office buildings, and more have become consumers of thermal imaging cameras. The more traditional markets, such as security, defense, and industry, are suffering from budget cuts, project cancellations, or postponements. Combine two of these elements, and the new EBT market is easily 3-4 times the size of the other markets combined.
But how effective are thermal cameras against the epidemic?
The jury is out on this, and opinions can vary from miraculous solutions to blatant fraud. The truth, as always, is something in the middle.
Let’s discuss some misconceptions:
1.Can thermal cameras detect viruses?
The answer is NO. The best the camera can do is tell you if someone has a hotter skin temperature than others. There are many reasons for an elevated body temperature, which are not all health-related, such as exercise or even sitting in a warm environment without air-conditioning.
2.Are the cameras accurate?
The accuracy debate is a significant and controversial discussion with much misinformation running around. When discussing accuracy, there are two conversations:
The first consideration is the accuracy of the camera itself versus a blackbody. For those who don’t know, blackbodies are devices that can regulate temperature very accurately (although not all are equal) and have a high emissivity level, which means they are almost not affected by surrounding heat or energy. All thermal cameras are calibrated against blackbodies. Still, some manufacturers have been using them in their EBT solutions to give the camera a consistent temperature reference to which it can adjust.
However, this can create other problems regarding thermal imaging vs. thermometric – but we’ll get to that later. The accuracy of the camera in this discussion talks about the camera itself. The detector’s sensitivity, internal reflections, lens aperture, noise level, and the calibration process itself. Also, if you read the fine print, most manufacturers quote accuracy levels, which are valid only in a controlled or laboratory environment, for example, a room with a steady 25°C and a slow shift in temperature (not more than 1°C per hour). Most field conditions don’t allow this – so this low level of accuracy is challenging to replicate in practice.
The more relevant conversation focuses on the fact that we are not looking for COVID-19 in blackbodies. We are looking for it in humans. And the substance known as human skin acts very differently. Unfortunately, to date, no medical models are predicting how skin will behave in different scenarios. We don’t know what the external skin temperature of a man weighing x who was exposed for y minutes to direct or indirect sunlight. So, while the blackbody may be spot on, it has no bearing on a person’s temperature reading.
So, while we can improve the first issue, the second one is more complicated. One way to circumvent it is by using population statistical analysis and looking for the gradient between the healthy population (which does have existing medical models) to the people with a higher temperature, which are statistical anomalies for such a camera.
3.Is there a difference between people with different skin color?
No – skin, whether dark or light, has the same emissivity, thickness, and temperature transference. In a thermal image, you have no idea of the skin color of the individual. More scientific proof that all people are created equal.
4.Are all thermal cameras suitable for temperature readings?
There is a difference between a thermal camera and a thermometric camera. A thermal camera developed for security and defense are used to detect threats and give situational awareness. We don’t care that two trees with different temperatures will have different colors – we care about the person standing between them. We manipulate the image, so the viewer has a better understanding of what he sees.
With thermometric measurement (as in – thermal temperature reading), we do the exact opposite. We want accurate temperature readings for each pixel in our screen. A thermometric camera will go through a rigorous calibration together with the lens, which often takes longer. Why? Because we need to offset, in the calibration tables, minute pixel-sized blemishes in the detector and lens. Those blemishes would be invisible in a thermal image, but they can skew the temperature reading and produce inaccurate results. Regularly we see suppliers who use regular thermal cameras with blackbodies to auto adjust the temperature reading as described above. But if you take that same blackbody and move it a meter to one side, you may discover the camera suddenly registers a different temperature – as not all pixels have a uniform calibration.
5.Does it matter where we scan in humans?
Yes and no. Indeed, the eye’s inner canthus (the tear duct) is the most relevant external point with the best correlation to internal temperature. People looking at the inner canthus will manage to avoid a lot of the effects of ambient temperature on the skin. The tradeoff is that the inner canthus is a tiny area. Since most of the thermal cameras in the EBT market use a 17µm pixel pitch and the human canthus is only about 4-5 millimeters wide, you’d need a big lens (expensive) or to stand very close (and still).
Additionally, people would need to remove their glasses. Most of the world’s health organizations consider the difference between a healthy and sick individual to be 1.5° C (or 2.7° F). That change is consistent whether you’re looking at the tear duct, the forehead, or a mouth. Thus, the solutions that look at the gradient temperature (population-based solutions) are just as effective when including the ambient temperature on the skin of the population tested.
6.Do people need to stop in front of the camera?
Not necessarily. It depends on the speed of the camera and the temperature detection algorithm. Some cameras can detect people walking very quickly as they only need a few frames to detect the temperature. So, if the camera is 9Hz (typical for many EBT cameras), you’ll need three frames to get an alert – you can detect approximately three people per second. No need to stop.
7.Will the camera work outdoors?
Outdoor environments can be very harsh for a thermometric camera, and most cameras will suffer from false alarms and misses. Some cameras have very advanced compensation algorithms for this, but they can’t take into account all the dynamic temperature changes, humidity, sporadic energy readings, and turbulence. Therefore the conditions can strain even the most advanced algorithm. Remember, as above – most EBT cameras need a steady environment to control their accuracy.
If you’re still reading, you’re probably asking yourself – why would we ever invest in such flawed technology? The truth is that after all is said and done – it’s still useful.
The WHO states that while asymptomatic transmission exists, it’s much less contagious then symptomatic transmission. Some doctors claim that a person with a fever sheds the virus five times more aggressively than a person with no fever.
Another factor is that as summer transitions to fall, we have another threat: our good old friend, the common flu.
In the US, last winter alone, the CDC estimated 35.5 million cases of Influenza. With symptoms similar to the flu, pressure for testing for COVID-19 due to sheer panic will be an epidemic in itself. Think about how, when one person has symptoms of the disease today, the time it takes for them to get test results leaves anyone who was in contact with them in a state of dread until the results are received (even when negative). Thus, the need to detect fever-related respiratory illnesses will be even more critical.
In some countries, there are clear regulations today for businesses to screen individuals for fever as they come into the establishment. While you can have a person in the entrance with a contactless thermometer – let’s remember that he must stop people for a 5-second check each time they come in. That would cause long lines in many places with high traffic. And during testing, standing less than 2 meters from the individual would throw social distancing out the window. If or when he got sick, the next day, he would start endangering everyone else he checked.
It’s better to screen automatically and only use the IR thermometer in cases where an alert was triggered and needed to be verified.
We’ve also seen much use of IR tablets. While they are low cost, think about the fact that a person usually needs to stand very close (less than 1 meter) from the monitor to be caught by the camera. Thus, spreading his germs on the glass or plastic cover of the tablet while being screened. The next person will enjoy that germ cocktail delivered by all the people tested before.
Also, let’s not forget the importance of the feeling of security itself. Take, for example, the security guard at the entrance to a bank. Will he be able to stop a well-planned bank robbery? We know from many Bruce Willis movies that it’s not always the case. To give a more real-life example, all the terrorist attacks committed in or at the entrances to airports were against airports that had extensive security. Yet, would we give up the security in an airport if it’s not 100 percent effective? Of course not.
In conclusion – Thermal EBT cameras are important. They aren’t a miracle cure, and they won’t stop the spread of the virus. And one should be careful of false promises. But along with other solutions (most importantly – masks), they can help protect us during these times and allow the wounded global economy to rejuvenate.