The (non)sense of thermal cameras to spot COVID-19

Can the application of thermal cameras in access control help spotting COVID-19?

The (non)sense of thermal cameras to spot COVID-19

Thermal cameras, the magic solution for spotting COVID-19?

The technology

Thermal imaging cameras detect infrared radiation and measure the surface temperatures of people and objects, by measuring the temperature differences between objects. Although the technology was initially developed for other purposes, it can be used to sense elevated skin temperature (EST).

For more accurate results, thermal cameras need to be calibrated in a lab and recalibrated annually. Most cameras will also need to be set up with a blackbody. Blackbody simulators serve as an optical reference point to obtain more accurate thermal data, since they have a theoretical emissivity of 1, which means that it perfectly absorbs and radiates all thermal energy.

The distance between the object or person and the camera also affects the accuracy of the reading, just like the camera settings. It is crucial that thermal cameras are setup correctly to receive the best results.

The use of thermal cameras to spot COVID-19

Thermal cameras measure elevated skin temperature (EST). It is important to note that EST and body temperature, as an indication of fever, are two different metrics. Skin temperature can give an indication about an individual's body temperature, however it should be used with caution. Normal body temperatures range from 36.1°C to 37.2°C. Skin temperature is around 33°C, with a peak around the area or the eyes closest to the nose, which ranges from 34°C to 36°C in normal conditions. Elevated skin temperature, as measured by thermal cameras, can give a proxy of a person's body temperature, though it remains an indication.

Most industrial infrared cameras have a margin of error of ±2ºC, with some manufacturers claiming a 0.3°C margin. It is important to note that these figures in the spec sheets are based upon lab experiments, under stable conditions. In a field setting the margin of error of ±2ºC can be lowered when using calibrated cameras, a blackbody simulator and an ideal distance.

A raised body temperature of 38°C indicates a fever. With normal body temperatures, ranging from 36.1°C tot 37.2°C, the margin of error should be less than 0.8°C.

A false sense of security

Even under the assumption that thermal imaging cameras can detect elevated skin temperature as a proxy of body temperature with a margin of error <0.8°C, the technology faces other challenges according to Doctor S. Kerré (M.D). A person infected with COVID-19 is contagious 48 hours prior to having symptoms, like fever. This would result in a false negative screening. In addition, there are more situations that could result in false screening values. Some examples:

  • On a cold winter day, a person's temperature stays low for a couple of minutes, when entering a building. This would allow a person with fever to be incorrectly screened.
  • A person that bikes to work has an increased body temperature, which would result in a false positive screening.
  • Pregnancy can raise the body temperature by more than 0.5°C.
  • Wearing a face mask will also be likely to have an impact on skin temperature.
  • Studies have shown that body temperatures decrease in senior populations.
  • The intake of painkillers and antipyretics lowers the body temperature.
“Thermal cameras give a false sense of security” - Doctor S. Kerré (M.D.)
Thermal cameras have a margin of error of ±2°C

Conclusion

According to experts of the World Health Organization, thermal imaging cameras might be helpful for initial screenings. However, they are not sufficient for an actual diagnosis. They're simply not designed for medical purposes. Even under optimal circumstances, there will be a lot of false positive and false negative scans. False negative scans can form a huge threat, since it implies a false feeling of safety and might encourage risky behaviour.