Database Restricted Access

# Upper body thermal images and associated clinical data from a pilot cohort study of COVID-19

Tamez-Peña, J., Yala, A., Cardona, S., Ortiz-Lopez, R., & Trevino, V. (2021). Upper body thermal images and associated clinical data from a pilot cohort study of COVID-19 (version 1.0). PhysioNet. https://doi.org/10.13026/pgk4-gx55.

Goldberger, A., Amaral, L., Glass, L., Hausdorff, J., Ivanov, P. C., Mark, R., ... & Stanley, H. E. (2000). PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals. Circulation [Online]. 101 (23), pp. e215–e220.

## Abstract

The prospective upper body thermal images SARS-CoV2 association study was designed to test the hypothesis that thermal videos may aid in the early diagnosis of COVID-19. The study recorded a set of measurements from 252 participants regarding PCR results, demographics, vital signs, participant activities, medications, respiratory symptoms, and a thermal video session where the volunteers performed simple breath-hold in four different positions. The acquired data may be used to test clinical association questions regarding temperature patterns, demographics, and vital signs. Furthermore, it could be valuable to develop new computer algorithms for extracting useful scientific information from thermal videos.

## Background

COVID-19 is a respiratory disease caused by the coronavirus SARS-CoV-2 [1] that was declared a pandemic by the World Health Organization (WHO ) in March 2020. According to WHO and as of January 2020, the number of global cases reached 90 million, and the number of confirmed deaths reached two million. The respiratory illness may cause acute respiratory distress syndrome (ARDS) characterized by pulmonary infiltrates and hypoxemia, where dry cough, fever, and fatigue are the main symptoms [2,3].

The principal diagnostic tool for SARS-CoV-2 is a DNA test based on a PCR assay [3–5], which requires respiratory specimens extracted by nasal or pharyngeal swabs [4]. The results are typically delivered to patients between 2 and 5 days. Other approaches have been proposed [6], like using medical images, particularly computer tomography [7,8] with reported prediction accuracy of 89% and area under the receiving operative curve of 0.92. These results suggest that imaging may be an alternative diagnostic tool for COVID-19. Nevertheless, computed tomography (CT) uses ionizing radiation, requires unique installations along with a complicated process limiting the number of possible tests per equipment, and the economic costs can be prohibitively high for screening a large population.

Thermal imaging is a technology that registers the infrared light emitted by any warm object. The captured infrared light spectrum is analyzed and used to estimate, depict, and record the temperature of any surface point of the body [9]. Thermal cameras are a common technology used in industrial settings to detect equipment and infrastructure failures [10]. Thermal cameras are relatively cheap, can remotely register temperature, record videos, are easily operable, and digital images and videos can be analyzed using computer software. In medical settings, thermal imaging has been used to detect breast cancer tumors' location and the location of lung-fluid accumulation in pneumonia patients with acceptable accuracies [11]. Nevertheless, to our knowledge, thermal imaging and specifically thermal videos have not been comprehensively investigated as an alternative diagnostic tool of COVID-19. Infrared videos can be useful for COVID-19 detection because SARS-CoV-2 infection in viremia stages is characterized by body temperature changes and in breathing patterns [12]. Thus, in principle, video recording of body temperatures could be a powerful tool as an alternative, cheap, and massive screening method and detection at an early stage of the disease.

Here, we present the description of thermic imaging and other demographic and clinical records from a pilot study of 252 individuals who underwent the COVID-19 PCR-based test on their own and voluntarily participated in this study.

## Methods

### Ethics approval and consent to participate

The study was approved by the institutional review board code of the School of Medicine at the Tecnológico de Monterrey corresponding to study ID “P000402-TERMAL-COVID 19-CI-CR001 / Termal-COVID 19” entitled “Estudio Piloto del Poder Discriminante de los Patrones Termográficos Y Sintomáticos entre Pacientes Positivos de Covid-19 Contra Pacientes Negativos de Covid-19”. All participants provided written informed consent.

### Study Population and Design

The study cohort comprises volunteer subjects requesting a PCR test for SARS-CoV-2 at the Hospital Zambrano-Hellion of TecSalud in San Pedro Garza Garcia, Nuevo León, México. The inclusion criteria were individuals older than 18 years old requesting a PCR test and signed the informed consent. The exclusion criteria were individuals unable to hold a deep breath for at least 10 seconds or not willing to show their back's bare skin. Volunteers were recruited from July 22nd, 2020, to September 4th, 2020, generally between 9 am and 3 pm. During that period, 252 individuals were recruited. After signing the informed consent, we recorded each participating subject's primary demographic, current medication intake, early day eating and activity events, self-reported symptoms, vital signs, and a thermal video. Details are described in the following sections.

### Clinical and demographic data

The participant’s self-reported data included age, gender, weight, height, the possible source of SARS-CoV-2 exposure: close family, workplace, or school. We also included self-informed symptoms: fever, cough, sore throat, diarrhea, vomit, loss of smell, loss of taste, chills, headache, and muscle pain, or others. Record of early day consumptions: last meal type, alcohol, cigarettes, drugs, e-cigarettes, or vaping. Record of recent physical activity: resting, walking, running, jogging, gym. Record of drugs or medications taken in the last 24 hours: diabetes, hypertension, pain, fever, or others. For women, we recorded the last menstruation date. Vital signals records included: axillary temperature, blood pressure, heart rate, and oxygen saturation. Vital signs were obtained using the Welch-Allyn 71WT monitor. A survey template is included in this database.

### Thermal image technology

Thermal images were recorded in video mode, mostly at five frames per second, using a Digital Thermal Imaging Camera TI-128 from Omega Engineering Inc. (800 Connecticut Ave. Suite 5N01, Norwalk, CT 06854, USA, www.omega.com). The camera was connected via USB to a laptop computer running Windows® operative system from Microsoft® as suggested by provider instructions. Acquisition software Omega TI Analyzer version 4.1.8.6875 was used. Drivers and acquisition software were obtained and installed following provider instructions.

### Thermal image acquisition

All the study participants were informed about the procedure and how the thermal images look like, avoiding human visible light colors, face identification, and body details. Participants were instructed to follow a simple pose pattern to record upper body high-definition thermal image recording. Before recording, glasses, masks, caps, and t-shirts, were removed. Brassiere or clothes covering women's breast area, earrings, and necklaces were not removed. Participants were asked to pose with raised arms in front of the thermal camera, taking a breath and holding it for 10 seconds unless they felt uncomfortable or challenging to hold breath, then down arms. They were then asked to turn right (clockwise) 90 degrees 3 times to take images from the left side, back, and right side, also uprising arms and holding the breath for 10 seconds in each position. A schema is shown in Figure 1. The video imaging and the acquisition procedure lasted around 1 minute. A background snapshot image was also taken before or after each procedure to record the background temperature.

### Image pre-processing

Raw videos and background snapshots are included in this repository without any processing. In addition, we carefully split each video into four scenes corresponding to the front, left side, the back, and right side and converted them to MPEG4 format to facilitate their inclusion into pipelines to the image processing scientific community.

### COVID-19 PCR diagnosis

The access to PCR (polymerase chain reaction) report was confirmed by participants signing the informed consent. Nasopharyngeal swabs were taken by qualified personal and deposited in a vital transport medium following standard procedures. The PCR testing was performed by the Clinical Laboratory from the Hospital San José, also part of TecSalud facilities in Monterrey, Nuevo León, México. The PCR procedure involves commercial PCR tests such as the TaqMan 2019-nCoV Assay Kit v1 & v2 from ThermoFisher. The PCR result is positive if an amplification signal is detected; otherwise, a negative result is determined. The PCR diagnostic includes an estimation of the viral load from an internal positive control curve.

## Data Description

Table 1 shows the summary of the dataset containing participant information (4 fields), previous exposure (4 fields), vital signs (5 fields), symptoms (13 fields), last meal (2 fields), previous COVID test (2 fields), other consumptions (3 fields), recent physical activity (4 fields), medications (6 fields), the results from PCR (2 fields), and the thermal images (6 fields).

Table 1. Summary of the clinical and demographic data.

 Data section Fields Participant information ID, Age, Gender, Weight, Height, Last menstruation (LMP) SARS-CoV-2 Exposure House, Hospital/Clinics, School/Work, Other Vital signs Temperature, Blood pressure, Cardiac rate, O2 Saturation, Respiratory rate Symptoms (< 24 hours) Fever, Cough, Throat pain, Diarrhea, Vomit, Smell loss, Taste loss, Shivers/chills, Headache, Myalgia, Generalized Arthralgias, Others Last meal Time, Type Previous COVID test Results Other consumptions (< 24 hours) Alcohol, Tobacco, Drugs/Vaping Last physical activity (< 2 hours) Resting, Walk, Jogging, Gym Medications (< 24 hours) Diabetes, Hypertension, Pain, Fever, Others, Which Imaging procedure Local Date, Time, Room Temperature Imaging quality summary Observations such as mask, necklace, beard, mustache, hair, blurriness, focus, and others PCR Diagnosis* Result, (Estimated viral load)

* Not defined by the participant, but by laboratory results and/or image analysis.

The image data files, described in Table 2, consist of original raw files from the camera software (.IRS files), background images just before or after taking the thermal video, four files from each position to be easily processed and taken for imaging analysis pipelines (see the Technical Validation section), and a pre-processing script.

Table 2. Summary of the video recordings and files.

 Folder or File Content *.IRS file Raw thermographic video as recorded and exported by the camera. One file per participant donor () Available through a full data request. See Usage Notes. *.mp4 Gray level complete thermographic video in MP4 format (1024x768 pixels) for each donor (),. Available through a full data request. See Usage Notes. ./termal_mpg_Data//Front/Front.mp4 ./termal_mpg_Data//Back/Back.mp4 Video segment representing the specific position in MP4 format (1024x768 pixels). can be Back, Front. Left and Right Views available through a full data request. See Usage Notes. ./termal_avi_data/PreProcessSubject.m Subject-specific Matlab script that generated the AVI files. The script indicates if the image will be deblurred, and/or flipped 90 degrees. It calls: GenerateAvi_front.m, GenerateAvi_back.m; ./termal_avi_data//Front/ClearFront.avi ./termal_avi_data//Back/ClearBack.avi Video segment representing the specific position in standardized AVI format. can be Back, Front. Left and Right Views available through a full data request. See Usage Notes background_jpg/

### Patient Privacy

Time Stamp (Procedure Date Key): The data and acquisition times were encrypted to avoid deidentification through involuntary access to hospital records. Patients and background JPG images contain the same date/time key procedure.

Original Files: The original video files (*.IRS) may be used to deidentify participant subjects. Hence these files were not uploaded into the repository. Furthermore, the left and right views may be used to deidentify study participants, hence those files were not uploaded.

## Usage Notes

Researchers may use the data to explore methods for the detection of:

• COVID-19 through thermal images.
• Subjects affected by respiratory issues
• Association of thermal images and vital signs, among others

The community is free to explore many more research ideas. The release includes standardized *.avi files for simple exploration of those ideas. Researchers may also choose to work with the *.mp4 files to explore different approaches for data standardization and/or image restoration.

### Data Quality Issues

Several subjects were captured out of focus and/or in landscape orientation. The subject_description.csv` file contains a detailed description of these quality issues in the "Thermal Quality Observations" column. Other quality notes are camera motion, mask, beard, mustache, necklace, long hair, and eyeglasses.

### Preprocessing

The AVI files are processed files of the MPG4 files. The AVI files are standardized versions of the MPG4 files. Standardization procedures were:

• Cropping to remove the grayscale color bar.
• Standardized frame rate (5 fps),
• Background Temperature stabilization,
• If requested: deblurring, and/or landscape orientation correction.

MATLAB R2021 with the image processing toolbox was used to generate the AVI files. PreProcessSubject.m file in each subject folder specifies if a specific subject was deblurred or flipped. The database contains all the MATLAB scripts required to generate the AVI files from the MPG4 files.

### Full Data Request

Interested researchers aimed to detect COVID-19 from those files are encouraged to contact the Corresponding Author with study proposals. Once the proposal is reviewed and approved, it may be possible to privately share original source data.

## Release Notes

An earlier version of the dataset was made available on Figshare [13].

## Acknowledgements

We thank the staff from Hospital Zambrano-Hellion, especially Ricardo Marroquin, Adrian Flores, Myriam Madelon Marcos, and Laura Garcia, for providing kind assistance during the project development. We thank Carolina Tamez-Gonzalez and Patricia Gonzalez-Cerna for transcribing and correcting clinical databases. We thank Ignacio Fuentes for organizing the transport of cameras. We also thank Andrea Celis-Terán for the artistic 3D scene representation of the body and video acquisition.

## Conflicts of Interest

The authors have no conflicts of interest to declare.

## References

1. Wu, F. et al. A new coronavirus associated with human respiratory disease in China. Nature (2020) doi:10.1038/s41586-020-2008-3.
2. Zhou, P. et al. A pneumonia outbreak associated with a new coronavirus of probable bat origin. Nature (2020) doi:10.1038/s41586-020-2012-7.
3. Guan, W. et al. Clinical characteristics of coronavirus disease 2019 in China. N. Engl. J. Med. (2020) doi:10.1056/NEJMoa2002032.
4. Huang, C. et al. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Lancet (2020) doi:10.1016/S0140-6736(20)30183-5.
5. World Health Organization. Coronavirus disease (COVID-19) technical guidance: laboratory testing for 2019-nCoV in humans. https://www.who.int/emergencies/diseases/novel-coronavirus-2019/technical-guidance-publications
6. Zhai, P. et al. The epidemiology, diagnosis and treatment of COVID-19. Int. J. Antimicrob. Agents (2020) doi:10.1016/j.ijantimicag.2020.105955.
7. Wang, S. et al. A deep learning algorithm using CT images to screen for Corona Virus Disease (COVID-19). medRxiv 2020.02.14.20023028 (2020) doi:10.1101/2020.02.14.20023028
8. Mei, X. et al. Artificial intelligence–enabled rapid diagnosis of patients with COVID-19. Nat. Med. (2020) doi:10.1038/s41591-020-0931-3
9. Ring, E. F. J. & Ammer, K. Infrared thermal imaging in medicine. Physiological Measurement (2012) doi:10.1088/0967-3334/33/3/R33.
10. Foudazi, A., Edwards, C. A., Ghasr, M. T. & Donnell, K. M. Active Microwave Thermography for Defect Detection of CFRP-Strengthened Cement-Based Materials. IEEE Trans. Instrum. Meas. (2016) doi:10.1109/TIM.2016.2596080
11. Arora, N. et al. Effectiveness of a noninvasive digital infrared thermal imaging system in the detection of breast cancer. Am. J. Surg. (2008) doi:10.1016/j.amjsurg.2008.06.015
12. Evertsen, J., Baumgardner, D. J., Regnery, A. & Banerjee, I. Diagnosis and management of pneumonia and bronchitis in outpatient primary care practices. Primary Care Respiratory Journal (2010) doi:10.4104/pcrj.2010.00024
13. Tamez-Peña, Jose; Trevino-Ferrer, Andrea; Díaz-Guerra, Carlos; Ledesma-Hernández, Meritxell; Esparza-Sandoval, Alejandra Celina; Yala, Adam; et al. (2021): ThermalCOVID19. figshare. Dataset. https://doi.org/10.6084/m9.figshare.13296887.v3

##### Access

Access Policy:
Only registered users who sign the specified data use agreement can access the files.