Non-contact portable infrared thermometer for rapid influenza screening

Improvements in transportation technology combined with the growth and human population mobility have brought spreading rate of infectious diseases to new level. Fever is most common indicator for many infectious diseases-particularly influenza- and can be easily screened using infrared thermography. However, most of thermography systems are expensive and non-portable. In worse case, pandemic flu outbreak in densely populated area will probably cause long queue of infected persons in crisis centre. To address those issues, we designed a non-contact portable infrared thermometer for fever screening in clinical settings. The device is also equipped with intuitive means to input symptoms associated with pandemic flu. The body temperature reading and list of symptoms is printed in ticket - size summary. Each suspect will carry this summary when seeing doctor for further examination. Inter - device testing under simulated clinical conditions and statistical analysis on temperature measurement between 35 to 38 °C indicates that overall bias and precision was 0.2 °C and ±0.4 °C respectively. A single complete process may be finished in around one minute.

SECTION I. Introduction

Current improvements in global transportation infrastructure combined with the growth and movement of human populations make transmission of infectious diseases possible almost anywhere in the world only in a matter of hours [1]. To delay or break the chain of transmission, rapid mass screening methods have been incorporated [2]–​[4]. Rapid fever identification of individuals suspected infectious diseases is an important component of the efforts because fever is a common indicator of many infectious diseases. This kind of screening was used by most of affected countries during the severe acute respiratory syndrome (SARS) outbreak in 2003 and the influenza A pandemic (H1N1) outbreak in 2009 [5]–​[8].

Most of the cases of SARS/H1N1 have occured among travelers returning to their hometown from other parts of the world, who contracted those diseases while abroad. This recognition should become a concern to university officials because of the high volume of faculty, students, and visitors traveling to and from affected areas, and because of the potential for rapid transmission in the crowded campus setting.

Simple, rapid, and inexpensive diagnostic testing may be utilized when appropriate to assist healthcare practitioner at student health centre to perform efficient screening for suspected individuals. Infrared thermography is a non-contact and fast method of monitoring body temperature. For these reasons, it has been used extensively for fever screening in many countries during the outbreak [9]–​[10]. However, most of thermography systems are expensive and non-portable. In this regard, we designed a non-contact portable infrared thermometer for fever screening of individuals with suspected infectious diseases in clinical settings. The device is not merely a thermometer but it also equipped with intuitive means to input symptoms associated with the pandemic influenza. The acquired body temperature together with the confirmed symptoms are then summarized and printed using a mini thermal printer. Each individual will carry the screening summary when seeing the doctor for further examination. This scenario is expected to save time that is normally wasted unnecessarily during interview between doctor and patient since it has been done in reception desk based on predetermined questions.

SECTION II. Theoretical Background

A. Principle of Non — Contact Infrared Temperature Sensing

All matter emits energy in form of infrared radiation. Infrared wavelengths are usually expressed in microns, with the infrared spectrum extending from 0.7 to 1000 microns. In practice, 0.7 to 14 micron band is used for temperature measurement. Infrared thermometers typically use a lens to focus infrared radiation from an object onto a sensor called thermopile. The thermopile absorbs the infrared radiation and turns it into heat. This heat is eventually converted into electricity for further processing.

A solid understanding of infrared technology and its principles results in accurate and precision temperature measurement. There are several factors that determine accurate measurement. The most important factors are emissivity, field of view, distance to spot size, and location of the hot-spot. Definition of each factor is detailed below. Fig. 1 may facilitate better understanding upon the descriptions.

• Emissivity: All objects reflect, transmit, and emit energy. Only the emitted energy indicates the temperature of the object. Measurement errors are often caused by infrared energy being reflected or transmitted by light sources. For most organic materials and painted or oxidized surfaces the emissivity is 0.95. If a thermometer with fixed emissivity is going to be used to measure the surface temperature of a shiny object, a certain compensation technique must be employed.

• Field of view: The sensor measures average temperature of all objects in the field of view (fov) of the sensor. The fov is given in degrees of the arc drawn around the normal to the center of the spot. The narrower the fov, the better the accuracy of temperature reading. The FOV may be expressed as $$\begin{equation*} fov^{0}=2\times((tan^{-1}(d/f)\times 180/pi)\tag{1} \end{equation*}$$ where f is the distance to the target and d is the diameter of the spot. Suppose an infrared thermometer using detector which has fov of 5° the circular spot from 23 cm away is 1 cm in diameter. It has to be ensured that the target is larger than the spot size the unit is measuring. The smaller the target, the closer the measurement should be made. The target should be at least twice as large as the spot size to guarantee high accuracy.

• Distance to spot (d:s) ratio: This ratio is the size of the area being evaluated by infrared thermometer in relation with the distance. As the distance increases, the area being measured becomes larger. The larger the ratio number, the better the thermometer resolution and the smaller area that can be measured. The laser sighting included in some instruments to help better aiming of the spot. For example, if a measurement would be made on a target with diameter of 5 cm and the infrared thermometer has a d:s ratio of 8:1, the maximum distance at which the temperature can be reliably measured is 40 cm. Beyond this distance, whatever else falls within the spot is also being measured.

Fig. 1.

Factors affecting temperature measurement accuracy using infrared thermometer.

B. Contact vs. Non-Contact Body Temperature Measurement

Temperature is one of the vital signs measured using electronic contact thermometers (rectal, oral, axillary), chemical thermometer (axillary, forehead), or infrared thermometer (tympanic, temporal artery). The mercury-in-glass thermometer has been used as the standart for human temperature measurement for hundreds of years, but it is going to be recalled from the market because it is potentially toxic [11], [12]. Axillary thermometers are less invasive than oral or rectal thermometers but required to be fixed closely in the axilla for the certain time to acquire an accurate measurement, approximately 30 seconds or longer [13]. Infrared tympanic thermometers may provide a more convenient means of measuring temperature, as the ear is readily accessible and readings may be acquired within seconds [9]. However, a review comparing infrared tympanic thermometry to rectal thermometry reported poor agreement between these two methods. This heterogeneity may be caused by presence of ear wax and the curvature of the ear canal may make it difficult to reach the tympanic membrane [14].

Non - contact infrared thermometers can be used to measure human body temperature rapidly and non - invasively. They can provide valid readings within seconds. Most non - contact infrared thermometers measure over the central forehead area, but temperature over other body surfaces may also be measured. Since the use of non - contact infrared thermometer does not involve any body surface contact, the risk of cross - infection is negligible and neither disinfection is required. Non - contact infrared thermometer temperature readings correlated strongly with rectal temperatures measured using mercury-in-glass thermometers with mean difference was less than 0,1 °C [15]. However, the infrared thermometers used in the studies were commercial products which surely incorporated certain compensation technique to match results from conventional reading methods. This hypothesis is underpinned by a study that developed thermopile array for measuring body temperature using low-cost thermopile detectors [16]. Comparison of the maximum facial temperature and reference axillary temperature showed an average mean difference of 1.03 °C. This indicates that a correction factor should be included when trying to acquire core body temperature via surface readings.

Another studies indicate that forehead body temperature may not be an accurate predictor because it is could be affected by many variables, primarily environmental temperatures [17], [18]. In order to obtain good measurement, subject is suggested to rest for around 5 minutes in the health station before any measurement is made [19].

C. Influenza Case Definition [20]

The case definition of influenza is sub-categorized into probable and confirmed cases. In view of the time required for laboratory confirmation of pandemic influenza infection, the probable case definition will be the working definition for operational considerations and the contact definition has been developed in relation to this.

1) Probable Case

Persons are considered probable pandemic influenza cases when following conditions are fulfilled.

1. Abrupt onset of fever more than or equal to 38°C (except in persons aged 60 years and above); and

2. non - productive cough; and either a positive epidemiological link or a positive rapid test kit result if available.

3. Fever may often be absent in persons aged 60 years and above. Therefore, in the absence of fever, any of the following symptoms, in addition to non - productive cough, should raise a high index of suspicion for persons in this age group: malaise, chills, headache, and myalgia.

2) Confirmed Case

Persons are considered confirmed pandemic influenza cases when there is laboratory confirmation of infection with pandemic influenza.

SECTION III. Material and Methods

In the event of an pandemic flu outbreak in densely populated area, there is a chance of long queue of infected persons in the health clinic or crisis centre. Generally, such facilities have reception desk where incoming pre - registered patients obtain queue ticket. For new patients, a new medical record will also be provided in addition to the ticket. In some facilities, patients will have blood pressure and temperature check on reception desk.

An effective and efficient registration and initial check method is required to handle extraordinary situations. During the height of epidemic, separated reception mechanism for old and new patients is no longer relevant as well as time consuming and risky contact screening.

First, each individual will have infrared body thermometry measuring the forehead body temperature to detect fever. After that, each suspect will have brief interview about flu - related symptoms that should raise a high index of suspicion. The list of the questions have been developed based on a response plan [20] and programmed in the device. The answers are stored in the device's internal memory. Finally, the device will generate a printed summary containing body temperature reading and list of confirmed/unconfirmed symptoms. When seeing the doctor for further examination, the patient will show the summary so hopefully the time spent for diagnosis should be shorter. This process is illustrated in Fig. 2.

Fig. 2.

Proposed method to handle flu outbreak.

A system was designed to obtain valid body temperature via forehead measurement and to input influenza symptoms based on interview between physician and patient. The system is based on a microcontroller, with sensors and peripherals wired to and controlled by the microcontroller. Components were selected to provide required functions, integrate easily with the microcontroller, were reasonable in cost and available off the shelf, and battery operated.

A. System Overview

The primary components of the system are an AVR Atmega8A microcontroller, infrared thermometer, ultrasonic range sensor, graphical lcd, analog joystick, voltage regulators, and thermal printer. A block diagram of complete system is shown in Fig. 3.

An infrared thermometer, MLX90614 from Melexis is incorporated to acquire body temperature. The MLX90614 consists of infrared-sensitive thermopile detector, signal conditioning chip, 17-bit adc, and powerful dsp in single unit. An optical filter is added onto the package to make the thermometer immune against ambient light and sunlight. The wavelength pass-band of this optical filter is from 5.5 to $14 \mu\text{m}$ range. The microcontroller communicates with the thermometer using I²C communication protocol with the speed of 30 kHz. The standard version has accuracy of ±0.5 °C and resolution of 0.02 °C for object temperature of 0 to 60 “C and ambient temperature of 0 to 50°C. Using this version which also has fov of 90°, to measure temperature of target with spot size of 5 cm in diameter, the distance between thermometer and forehead should be between 2.5 to 5 cm for best accuracy. The device can operate normally using power supply voltage of 2,6 to 3,6 V with current consumption up to 2.5 mA.

Fig. 3.

Block diagram of the system.

An ultrasonic range sensor from DT-Sense is employed to ensure that temperature measurements are performed within proper distance. This sensor integrates a pair of ultrasonic transducers, amplifier, and controller in a single board. The range within the sensor can reliably measure is 2 cm to 3 m with accuracy of ±0.5 mm. The device connects to microcontroller using general input - output pin. The supply voltage requirement is 5 V and it consumes 17 mA when measuring distance.

The user interfaces with the system using a display commonly found in old Nokia 5110 cell phones and a two - axis analog joystick used by most game controllers. The display is a $84\times 48$ pixel monochrome lcd with diameter of 1.6 inch. It is small, but very readable and come with led backlight. The lcd interfaces to microcontroller through a spi interface with the speed of 250 kHz. Operation voltage is between 2.7 to 3.3 V with maximum current consumption of 15 mA. The use of analog joystick is intended for quick menu navigation mainly during symptoms inputting. The x - axis is used to jump to next predefined symptom while the y - axis is used to select whether a symptom is confirmed or unconfirmed. The center push - button in the joystick is used to perform one - touch temperature measurement. It takes two adc channels to track down joystick movement on both directions. In addition to display and joystick, a buzzer is incorporated to indicate that valid body temperature has been acquired.

A battery pack, consisting of two 3.7 V lithium-ion, 1800 mAH batteries provides unregulated power supply for the system. Most of the components in the system are designed to operate from 3.3 V supply, and an AMS1117-3.3 voltage regulator is used to adjust the battery voltage to a stable 3.3 V supply. Another regulator based on similar chip is used to provide 5 V regulated supply for ultrasonic range sensor and buzzer. Fully charged battery pack will produce about 8.4 V, while when depleted its voltage may be closer 6.4 V. This voltage variation is going to be monitored with the help of adc with reference voltage of 3.3 V therefore a simple voltage divider is employed to bring down the battery voltage to a range between 2.1 to 2.6 V.

The thermal printer comes in compact size and takes 57 mm wide thermal paper. It will print strings passed to it via serial TTL with default 19200 bps baud rate. Printing speed is between 5 to 8 cm/s. The supply voltage requirement is quite flexible, 5 to 9 V and it draws approximately 2 A when printing.

B. System Operation

System operation is governed by firmware inside the microcontroller written in C language. The firmware enables microcontroller to communicate with peripherals attached to it.

From user's point of view, system operation can be partly described in Fig. 4 and Fig. 5. First, user is required to perform temperature measurement by pressing the joystick. If the device is no more than 3.3 cm away from the forehead, the infrared sensor access routine will be executed and temperature reading will be displayed. Otherwise an out of range icon will show; in this case user should move the device closer to forehead so that valid measurement can be obtained. The fever warning is set to 38°C so any readings equal or beyond that point will cause fever icon to show. To repeat temperature measurement, user should move joystick to the left while moving joystick to the right causes the sequence to proceed to enter age category sub - menu.

Age category is divided into two groups, below or equal 60 years old and above 60 years old. User may move joystick up or down to switch between age categories. To proceed to first symptom sub - menu, user needs to move the joystick to the right. In this sub - menu, user may move the joystick up or down to switch between yes or no choice. The choice should be selected based on the suspect's answer upon the first symptom. Moving joystick to the right automatically saves current choice and proceeds to the next symptoms. The procedure repeats until the sixth symptom sub - menu is completed. A ticket - size summary will be disposed from the thermal printer as soon as the user moves the joystick to the right at sixth symptom sub - menu. Screen is now back to the initial menu.

Fig. 4.

Flowchart for temperature reading routine.

Fig. 5.

Flowchart for entering last symptom.

SECTION IV. Results and Discussion

A prototype of non - contact thermometer has been successfully designed, built, and tested in laboratory. This thermometer is equipped with feature to enter symptoms and can be connected to thermal printer. It is also portable and powered by rechargeable battery with huge capacity, mostly to provide sufficient power for the thermal printer. In accordance with the function, this instrument is called “Termoprint” and is illustrated in Fig. 6.

Fig. 6.

The termoprint, infrared thermometer combined with thermal printer.

The accuracy and precision of the prototype has been verified relative to the digital thermometer Signstek 6802 II under simulated clinical conditions. This reference instrument employs type-K thermocouple as temperature sensor and has accuracy of ± 0.1 °C. The measurement object is $10\ \text{cm}\times 30$ cm metal plate which is bolted on two $20\ \Omega$/20 W connected in parallel. A voltage source from laboratory power supply is then applied to the circuit. A variety of forehead temperature were simulated by varying the output voltage of the power supply. The plate will be warmer as the voltage increased and vice versa.

The temperature sensor of reference instrument is fixed on the surface of the metal plate while Termoprint senses temperature of metal plate from the distance approximately 3 cm away. Fig. 7 illustrates the testing method.

Fig. 7.

Testing setup to verify performance of termoprint.

Voltage output from laboratory power supply was varied from 3 to 3.5 V to obtain temperature variation between 35 to 38 °C on the plate. A certain amount of time delay should be taken into account when switching between voltages to allow temperature of metal plate to be steady. The measurement were recorded and analyzed on three steady points indicated by the reference instrument. Each point were captured 20 times by the prototype to facilitate statistical analysis upon the data.

Inter - device testing and statistical analysis on temperature measurement between 35°C–38°C indicates that overall bias and precision was 0.2 °C and ±0.4 °C respectively. The reference instrument has smaller deviation between each measurement because its sensor was tightly attached on the object while measurement performed by the prototype is non - contact by nature and it is affected by a number of environmental variables. The degree of accuracy and measurement range is also comparable to similar commercial product such as Thermofocus [21].

A scatterplot between Signstek 6802 II and Termoprint with the line of identity is shown in Fig. 8. The scatter diagram indicates strong correlation between two variables with $\text{r}^{2}=0,94$. Fig. 9 shows a moment when Signstek 6802 II and Termoprint were taking temperature reading on the object at certain temperature. The visual warning (“HI”) appears on the display since the temperature reading exceeds predetermined fever cut off point of Termoprint. The prototype compared well to the reference instrument during this laboratory simulated test.

Fig. 8.

Temperature scatterplot between signstek 6802 II and termoprint.

Fig. 9.

Termoprint and reference instrument taking temperature reading simultaneously.

Based on simulated tests in the laboratory, the symptoms registration may also performed in very fast and convenient way. This feature can be achieved mostly because Termoprint incorporates two-axis analog joystick to navigate through the menu and thermal printer to print the assessment summary. A single complete process may be finished in around one minute. The summary is printed at the end of assessment in form of a ticket containing current date and time, temperature reading, age category, and the list of conditions that is absent or present in a suspect based on interview. Fig. 10 shows an example of a assessment ticket.

Fig. 10.

Example of an assessment ticket.

SECTION V. Conclusions and Future Works

Termoprint exhibits comparable performance to reference instrument in terms of accuracy and precision particularly in temperature range between 35 to 38°C and measurement distance up to 3 cm. The performance, in terms of accuracy and measurement range is comparable to similar commercial product. The utilisation of analog joystick and thermal printer contributes significantly to fast menu navigation and assessment summary generation. Results like these pave a way to the development of medical grade system to perform rapid screening in case of influenza outbreaks.

Possible future works includes replacement the current temperature sensor with the medical grade one to get equal accuracy in longer measurement range. In the firmware side, it is envisioned to incorporate specific algorithm to compensate forehead temperature reading so it will actually reflect core body temperature.

ACKNOWLEDGMENT

The authors would like to acknowledge faculty members at the Department of Electronic and Computer Engineering and Office of Vice Rector for Research and Service to Community at Satya Wacana Christian University for their supports of this research.