World IEC Fever Screening Standards Explained
While 'fever detection' is new to most people, the technology has been practiced for many years and has a series of international standards that we explain herein.
In particular, the world's two top international standards groups, the IEC and ISO, have published 3 standards covering fever (i.e., febrile) screening:
- IEC 80601-2-59:2017 Medical electrical equipment — Part 2-59: Particular requirements for the basic safety and essential performance of screening thermographs for human febrile temperature screening
- ISO/TR 13154:2017 Medical electrical equipment — Deployment, implementation and operational guidelines for identifying febrile humans using a screening thermograph
- ISO 80601-2-56:2017 Medical electrical equipment — Part 2-56: Particular requirements for basic safety and essential performance of clinical thermometers for body temperature measurement
We purchased those 3 standards for 530 Swiss Francs (~$550 USD), studied them and prepared this guide to explain how they apply to fever sceening, including:
- Focus Area: Inner Eye, Other Areas "Unreliable"
- ISO Face Examples Show Importance of Inner Eye
- Eyeglasses, Hair Shouldn't Obstruct Face
- Impact of Obstructions Shown by ISO
- Manufacturers Contradict Standards
- 'Face Detection' AI and Temperature Algorithms
- Minimum Lab Accuracy Level
- Faces Must Be Parallel to Camera
- Minimum Face Pixel Requirements
- Secondary Measurement Required
- Clinical Thermometer Requirements
- Ambient Temperature Must Be Controlled, Ensure No One Is Sweating
- Avoid Bright Lights, A/C Drafts, Other Heat/Cold Sources
- Secondary Screening Area Requirements
- Fever Cams For Use "Under Indoor Environmental Conditions"
- Long List of Facilities Recommended
- Ease-of-Extraction Emphasized
- No Toilets Nearby
- Backdrop Requirements: Non-Reflective, Avoid Dark Colors
- People Should Be Screened One-at-a-Time
- Data Should be Kept For One Month/GDPR Impact
- Applicability / Legality
The International Electrotechnical Commission (IEC), based in Geneva, Switzerland, is the world's top standards organization "for all electrical, electronic and related technologies" with 85 member countries, including the US, PRC (China), France, Germany, India, Japan, etc.
The International Standards Organization (ISO), also Geneva-based, publishes international standards but for a broader variety of products, and includes 164 member countries, including the nations mentioned above.
Both organizations have published detailed standard guides for fever detecting thermal camera solutions, and the FDA has endorsed them, including in their most recent guidance stating:
[cameras are] tested and labeled consistent with the following standard: IEC 80601-2-59:2017
[the camera's] labeling references and is consistent with the guidelines in ISO/TR 13154: 2017
Thus, in this guide, we use both organizations' recommendations, as the two groups explicitly state they collaborate "closely" and recommend each other:
IEC collaborates closely with the International Organization for Standardization (ISO) in accordance with conditions determined by agreement between the two organizations [from IEC 80601-2-59:2017]
ISO collaborates closely with the International Electrotechnical Commission (IEC) on all matters of electrotechnical standardization [...] IEC 80601-2-59 provides recommendations on the types of temperature screening operations and relevant performance requirements for responsible organizations to adopt for their intended operation [from ISO/TR 13154:2017]
Focus on Inner Eye, Other Areas "Unreliable"
The IEC states the inner eye is the best area for body temp readings due to it being over an important artery, stating other body areas are "unreliable":
Facial thermography of surface areas other than the region medially adjacent to the inner canthi is unreliable, and may be complicated by perspiration, facial skin flushed from exertion, etc. The current evidence indicates that the region medially adjacent to the inner canthi is the preferred site for fever screening due to the stability of that measurement site. This is because this region is directly over the internal carotid artery.
ISO Face Examples Show Importance of Inner Eye
The importance of the inner eye area is best shown by a real-life thermal cam images from the ISO/TR 13154:2017. The figure below shows a man without a fever (afebrile) with a hit of 36.6°C on thermal camera:
For a man with a fever, his left eye shows 37.9°C and his right 38.4°C:
Both these thermal cam readings used a blackbody of 35°C, ISO notes.
Eyeglasses, Hair Shouldn't Obstruct Face
The IEC also states hair, eyeglasses, and other objects should not be covering someone's face:
Instruction to the OPERATOR to ensure that the FACE is unobstructed by hair, eyeglasses, and other objects because their presence will interfere with the ability of a SCREENING THERMOGRAPH to detect a febrile condition.
Impact of Obstructions Shown by ISO
The impact of obstructions is clear. Below is an ISO thermogram of someone wearing glasses, almost totally obscuring the critical inner eye area:
Even a hat has similar effects:
Notably, ISO states a face mask is also an obstruction, and leads to warmer-than-usual readings due to warm breath exhalations being reflected back onto the face:
The ISO notes it has adjusted the celsius display scale in the 3 images above to "demonstrate the thermal pattern more clearly".
Manufacturers Contradict Official Advice
However, many firms touting fever cameras do not disclose that glasses and masks severely impact accuracy, along with not disclosing the critical importance of inner eye readings.
Demo videos often show 'accurate' readings for people with glasses that totally obscure the inner eye area, e.g., from the PRC's Nuctech:
Or this Hikvision demo:
Or from this Dahua video that has been viewed 140,000+ times:
However, now, Dahua is recommending a snaking or winding pattern where people queue to get their faces scanned, as shown in this video they provided to us embedded below:
Minimum Lab Accuracy Level
IEC states the minimum laboratory accuracy for a thermal camera is ±0,5°C in the 34°C to 39°C range:
The temperature LABORATORY ACCURACY of a SCREENING THERMOGRAPH, including the measurement uncertainty shall be less than or equal to an offset error of ±0,5°C over the range of at least 34°C to 39°C
Generally, this is also the accuracy touted by most fever cam manufacturers, with many (like Sunell, Dahua, FLIR, Hikvision) touting ±0,3°C and Mobotix even claiming ±0,1°C.
IPVM is still trying to verify what laboratories manufacturers are using to verify the accuracy of their accuracy specifications.
Faces Must Be Parallel to Camera
IEC states faces should be parallel to the camera:
The WORKABLE TARGET PLANE of the SCREENING THERMOGRAPH shall accommodate a FACE that is positioned 0,75 m to 2,2 m above the floor. This requirement may be met by moving the infrared camera. The plane of the lens of the infrared camera also should be parallel to the FACE and in line with the TARGET.
The ISO adds that if this is not achieved, errors of "up to 1 °C" are possible:
Unless the face and camera lens are correctly aligned, errors of up to 1 °C can be introduced. Such errors usually lower the temperature value measured.
Here is one example that the ISO considers a "properly positioned individual":
This goes counter to manufacturers which show significantly angled faces, e.g. this Dahua video:
Minimum Face Pixel Requirements
IEC states cameras must be parallel in order to maximize the number of pixels in the face image, which should be a minimum 240 by 180:
This [parallel] requirement is intended to ensure that there are sufficient IMAGE PIXELS in the thermogram of the FACE and TARGET to permit an accurate assessment. The minimum display of the WORKABLE TARGET PLANE shall be 320 IMAGE PIXELS by 240 IMAGE PIXELS. In NORMAL USE, the thermogram of the FACE shall fill at least 240 IMAGE PIXELS by 180 IMAGE PIXELS.
Many of the manufacturers IPVM are using 400x400 (e.g., Sunell's Panda Cam) or smaller 320x288 sensors, which means they could only comply with this when reading a single face, not 10 or more in a row or the 40+ shown in Sunell's marketing below:
If a face on camera needs to be 240 pixels wide and a real face is about six inches wide, effectively a 400-pixel camera should have a field of view no wider than 10 inches. However, Sunell is only getting around 50 pixels per face at best, a fifth of the IEC recommendation.
The practical issue is that with fewer pixels on target it is harder to get precise readings since the readings of adjacent hotter and colder parts of a face will blend in, reducing accuracy.
No Mention of 'Face Detection' AI And Temperature Adjusting Algorithms
The ISO/IEC standards make no mention of such AI or of anything else helping overcome these obstructions. This sets up an issue where manufacturers may argue these 2017 standards are out of date.
Some manufacturers have also touted "compensation algorithms" they claim automatically adjust for the (well-known) difference between face skin temperature and actual body temperature. However, the ISO recommends that this "small difference" between inner eye temp and body temp be accounted for by adjusting the "threshold temperature", i.e. the specific temperature at which the system alarms:
The screening thermograph measures the skin temperature of the region medially adjacent to the inner canthus and not a body core temperature; there is a small difference in temperature between these two sites. This difference should be accounted for in the selection of the threshold temperature.
We are in the middle of testing multiple systems using on-the-move and low pixel density and plan to update our reporting here as soon as possible.
Secondary Measurement Required, Clinical Thermometer Requirements
A high temp reading cannot be automatically considered a fever, and must be confirmed with a clinical thermometer, the IEC says:
All individuals suspected of being febrile with SCREENING THERMOGRAPHS should be confirmed with a secondary measurement using a clinical thermometer.
The clinical thermometer should conform to a separate standard, ISO 80601-2-56.
This standard includes the now-ubiquitous infrared thermometers held over the forehead; however, such devices must be tested in both a lab and on a "sufficiently large group of human subjects", unlike 'regular' body thermometers (e.g. ones held in the mouth) which require only lab testing.
The ISO recommends laboratory accuracy of ±0,3°C for clinical thermometers.
Fever Cams For Use "Under Indoor Environmental Conditions"
The ISO/TR 13154:2017 and IEC 80601-2-59:2017 standards specifically state that fever screening is deployed under indoor conditions:
[IEC] SCREENING THERMOGRAPHS have been used at ports-of-entry, ports-of exit and the entrances to buildings under indoor environmental conditions with the intention of separating febrile from afebrile individuals to help prevent the spread of communicable diseases
[ISO] this document provides general guidelines for the deployment, implementation and operation of a screening thermograph intended to be used for non-invasive febrile temperature screening of individuals under indoor environmental conditions to prevent the spread of infection.
However, this has not stopped some firms, like the PRC's Rokid, from touting dubious outdoor fever detection solutions:
Avoid Bright Lights, A/C Drafts, Other Heat/Cold Sources
The IEC says the type of lighting matters, as it could affect accuracy. Specifically, bright lights such as incandescent or halogen should be avoided:
The RESPONSIBLE ORGANIZATION needs to be aware of the type of lighting used at the screening area. Lighting such as incandescent, halogen, quartz tungsten halogen and other type of lamps that produce significant interference (heat) should be avoided. The area chosen for screening should have a non-reflective background and minimal reflected infrared radiation from the surroundings.
IEC recommends A/C drafts be diffused to ensure they are not blowing onto people and cooling them:
Drafts from sources such as air conditioning ducts can cause forced cooling or heating of the FACE and should be baffled or diffused to prevent airflow from blowing directly onto the PATIENT.
The ISO adds that "sun-facing windows, radiant heaters, or sources of cold (cold windows or outside walls" can also "interfere" with accurate readings and must be avoided as well.
By contrast, below is an image from a May 2, 2020, Canadian article showing a Hikvision system monitoring the glass door entrance to a liquor store:
Long List of Facilities Recommended
The ISO recommends a wide variety of facilities implement fever screenings during a "severe pandemic":
- entrances to hospitals and clinics, including emergency rooms;
- entrances to critical infrastructure facilities;
- entrances to workplaces;
- entrances to schools;
- entrances to government buildings, including police and fire stations;
- entrances to other communal locations;
- public transportation.
This is the main reason "fever cameras" have gone from a niche product to a Booming Multi-Billion Dollar Market.
Ambient Temperature Must Be Controlled, Ensure No One Is Sweating
Controlling ambient temperature is important, as overly hot/cold people will not give accurate results, particularly if they are sweating, ISO says:
individuals being screened should not be too cold or too hot and especially not sweating.
The ISO states that ideally, the temperature should be 20°C to 24°C (68°F and 75°F) and humidity from 10% to 50%:
The temperatures measured by a screening thermograph can be influenced when the individual being screened is sweating. Sweating thresholds can vary according to a person’s fitness level, environment of residence, length of adaptation and the relative humidity. When humidity is controlled, these effects are minimized. To produce consistent and reliable results of the temperature screening process, it is imperative that the screening thermograph be situated in a reserved stable indoor environment with a temperature range of 20 °C to 24 °C and relative humidity range from 10 % to 50 %. [emphasis added]
The ISO states one way to achieve such conditions would a be a special walk-through booth:
These conditions can best be achieved by creating a local, controlled environment. EXAMPLE A walk-through booth.
Secondary Screening Area Requirements
In order to prevent "cross-contamination" (febrile individuals in the line infecting others), the ISO recommends that a "secondary screening area" be set up "removed from the general traffic flow" for people who are being confirmed for fever:
The secondary screening area should be at a tangent to the screening thermograph area, but removed from the general traffic flow. Screening near the entrance of the facility prevents comingling.
The secondary screening area should be properly equipped with "masks, wipes, disinfectants".
They should also include a method for safely and privately sending febrile individuals to quarantine or the relevant health department:
The secondary screening area is a care area that should equipped with a clinical thermometer and accessories that comply with ISO 80601-2-56 and should be staffed by qualified medical personnel. The secondary screening area should be equipped with sanitation supplies, e.g. masks, wipes, disinfectants. To prevent cross-contamination, the secondary screening area should be positioned to allow patient removal from the facility or to quarantine with reasonable privacy and with minimum exposure to others (maintaining crosscontamination prevention).
It's important to make sure the screening area allows maximum "ease of extraction", i.e. people can be removed without creating bottlenecks in the flow of people passing through:
The workflow has to be designed in a manner that allows ease of extraction of suspected febrile individuals
No Toilets Nearby
The ISO says bathrooms should not be near screening areas.
They explain this is to reduce infection risk and stop people with fevers from washing their faces to fool the system:
Toilets should not be proximal to the screening thermograph area. This is to both inhibit potential cross-infection and to prevent facial washing (alteration of the thermal profile) immediately prior to entering the screening thermograph area.
Cold water will temporarily cool the face allowing a person to trick the fever screening that they do not have a fever:
Backdrop Requirements: Non-Reflective, Avoid Dark Colors
The ISO also has recommendations for the backdrop behind the person being screened, chiefly that it not be dark and that it's not reflective:
The backdrop behind the individual being screened and, where utilized, side screens should be — thermally uniform, — non-reflective in the IR spectrum, and — not dark in colour in the visible spectrum (closer to white than black).
People Should Be Screened One-at-a-Time
To get the best possible shot of someone's inner eye, ISO recommends a single file line, and that people should "stop and pause":
The individuals to be screened are channelled into single file and caused to stop or pause so that the screening thermograph can capture the region medially adjacent to the inner canthi temperature distribution one individual at a time. Measuring individuals one at a time facilitates the capture of a reliable thermogram and allows the determination of potentially febrile individuals requiring secondary screening.
However, ISO does state that at high-volume situations, the system can operate in "near real time".
To minimize disruption in high volume situations, the response time and throughput of the screening thermograph should be capable of operating in near real time for rapid and effective screening. This can necessitate that the screening thermograph be highly automated.
But in low-volume scenarios, it's still best to ask people to stand still:
when the response time and throughput is not as demanding, the temperature screening operation can require the individual to stand still at a set location in front of the camera of the screening thermograph while the temperature is being screened
Again, firms like Sunell and its Panda Cam openly flout this requirement, advertising a simultaneous screening of 20 or more people:
Data Should be Kept For One Month/GDPR Concerns
ISO recommends all the data pertaining to thermal temperature checks be kept for one month:
The responsible organization should retain this information for at least one month (normal maximum incubation time for known infectious diseases). The responsible organization should be prepared to maintain the data for longer periods when deemed necessary by the public health authorities and other organizations ensuring protection of public safety.
Note: technically, the GDPR does not apply to thermal camera readings, as it only deals with the "processing of personal data" i.e. data that can identify a specific person - which thermal readings cannot. However, if the end user records a person's name and/or face and marks her as febrile, then that becomes subject to GDPR as health information is considered a "special category" of personal data. (Article 9). This means such data should be handled very carefully, i.e. kept secure and only shared with health authorities. As ISO notes:
The collected data are electronic health information and thus subject to privacy and security laws in many jurisdictions. As such, access to the collected data should be restricted and access records and all metadata should be maintained according to local law.
Applicability / Legality
While the recognition of IEC / ISO standards is broad and global, they are not generally legally binding, i.e., one does not risk going to jail for violating. However, they are often cited, recommended, or called out by government agencies, e.g., the US FDA here. As such, going against these standards carries some risk that organizations should consider. Also, if one does go against them and coronavirus infections subsequently are traced to that facility, it could increase legal risk (i.e., the organization did not follow the IEC / ISO standard). On the other hand, in the middle of a pandemic, the probability increases that standards will be ignored, at least short term.
Vote / Poll
**** ** * ***** ******* *** very ****** ** **** ******** *****. While ******* **** * *** ************** on * ********** **** ***** ** was ********* **** *** ********'* ********* will ** *** ***** *********:
*** ******* **** ** ******** ****** - ** *******.******** ********** **** ********** **** *** additional ********* *** ** *** ******** on **** ****. **** ***** * handheld **** ** **** *** ************ and ** ***** ***.* ******* ********* they *** ********** ******.
**** ***** **** ** ***** ** a **** **** ***. ***** **** people ***** ******** **** ** ****** in **** ******* *** *** ******* is ******* ** **** ***** **** a ******* ******* **** ********.