Data and code from: Application of infrared thermography coupled with Eulerian Video Magnification to monitor vital signs in wild seals, from sandy beaches to icy shores
Data files
Mar 19, 2026 version files 141.74 KB
-
distance_data_CUT.csv
6.86 KB
-
field_IRT_stats_post_wind.csv
20.04 KB
-
field_IRT_stats.csv
23.67 KB
-
field_IRT_stats.R
73.06 KB
-
README.md
6.91 KB
-
residuals_HR.csv
3.94 KB
-
residuals_RR.csv
3.73 KB
-
wind_HR_success_prob.csv
1.58 KB
-
wind_RR_success_prob.csv
1.97 KB
Abstract
Metabolic rate is a critical indicator of animal health, with respiration and heart rate used as common proxies for energetic expenditure. Infrared thermography (IRT) was previously paired with Eulerian video magnification (EVM) to capture changes in temperature associated with exhalation and blood flow for accurate and non-invasive measurements of respiration and heart rate. These techniques have been tested in a range of taxa, but only in controlled settings at zoological institutions and with imaging conducted at short range (1 m distance). This study aims to validate IRT-EVM in wild populations with minimal disturbance and under variable environmental conditions, and to determine the maximum distance that animals can be imaged for accurate vital sign measurements. Infrared videos were taken of pinnipeds in two dramatically different environments, including northern elephant seals (Mirounga angustirostris, n = 12) in North America and Weddell seals (Leptonychotes weddellii, n = 23) in Antarctica. ‘True’ respiration and heart rates were collected simultaneously by visually monitoring movement of the ribcage/nostrils and stethoscope/ECG readings, respectively. Using IRT-EVM, respiration rate was extracted from 74.3 % of individuals while heart rate was measured in 93.3 % of individuals, both with high accuracy (mean absolute error, respiration rate: 1.3 br~pm; heart rate: 7.4 bpm). Infrared-derived measurements were resilient to individual characteristics (species, body mass, molt status, blubber thickness, age class) but were impacted by wind speed, animal movement, and ambient temperature. Signal noise introduced by wind could be filtered out to still yield accurate vital signs. Camera resolution directly influenced the distance that respiration and heart rates could be measured. In this study, we demonstrate that IRT-EVM is a powerful, non-invasive method for assessing vital signs in free-living pinnipeds. These imaging techniques are likely to be applicable towards study of other mammalian species, with appropriate validations for use in new field environments to ensure accuracy of IRT-EVM derived vital sign measurements is maintained. This work makes strides towards monitoring wildlife metabolic indices to evaluate impacts of intra-annual and longer-term environmental change on population health.
Dataset DOI: 10.5061/dryad.1ns1rn96d
Description of the data and file structure
File information
File name: README_IRT_field_data
Author: Caroline L Rzucidlo
Other contributors: Roxanne S Beltran, Patrick W Robinson, Erin Curry, Amy Klink, Allyson G Hindle, Michelle R Shero
Data for manuscript: Application of infrared thermography coupled with Eulerian Video Magnification to monitor vital signs in wild seals, from sandy beaches to icy shores
Contact information:
Caroline Rzucidlo
MIT-WHOI Joint Program in Oceanography
crzucidlo@whoi.edu crzucidlo13@gmail.com
These data were used to validate the use of infrared thermography (paired with Eulerian Video Magnification) to accurately and non-invasively measure vital rates in pinnipeds across environmental conditions.
Description of the data and file structure
These data were collected at Año Nuevo State Reserve, USA (April - May 2022) and Erebus Bay, Antarctica (October 2022 - January 2023).
See the methods in the manuscript for the full description of how data were generated and analyzed.
This dataset includes seven .csv files and one .R file. Raw videos from FLIR cameras (.csq files) are available upon request.
Files and variables
Details for field_IRT_stats.csv:
- includes all individual and environmental characteristics from all imaging sessions, even ones where no true respiration or heart rate were recorded
- n/a represent data that were not available
- Variables:
- animal_ID: ID associated with the individual’s flipper tag
- species: elephant or Weddell seal
- video: refers to the video number
- ambient_temp: air temperature when imaging, in Celsius
- humidity: humidity when imaging, in %
- wind_kt: wind speed in knots
- wind_kph: wind speed in kilometers per hour
- sex: M or F
- age_class: pup, weaned pup, juvenile, or adult
- mass_kg: mass of the animal in kilograms
- blubber_cm: blubber thickness at the axillary
- molt_status_percent: the percent of the fur that was molted at handling
- immobilized: status of immobilization, either immobilized or not immobilized (not_immob)
- distance from camera m: distance between the subject of the video and the camera, in meters
- video_qual: ranked on a scale from 0-7 (see Methods in manuscript)
- RR_success: yes or no (or blank if no ‘true’ RR to compare to), whether or not respiration rate was successfully measured using IRT-EVM
- true_RR_brpm: ‘true’ respiration measured via watching ribcage or nostril movement, in breaths per minute
- IRT_RR_brpm: respiration measured via infrared analysis, in breaths per minute
- absolute_error_RR: error between true and IRT-derived respiration rate
- percent_difference_RR: percent difference between true and IRT-derived respiration rate
- MAE_RR: mean absolute error between true and IRT-derived respiration rate
- RR_location: location on the animal’s body that was imaged to extract respiration rate
- HR_success: yes or no (or blank if no ‘true’ HR to compare to), whether or not heart rate was successfully measured using IRT-EVM
- true_HR_bpm: ‘true’ heart rate measured via stethoscope or ECG, in beats per minute
- absolute_error_HR: error between true and IRT-derived heart rate
- percent_difference_HR: percent difference between true and IRT-derived heart rate
- HR_location: location on the animal’s body that was imaged to extract heart rate
Details for field_IRT_stats_post_wind.csv:
- includes all the same variables and data as above (field_IRT_stats.csv), except these RR values are after EVM correction for wind interference.
Details for distance_data_CUT.csv:
- only includes animals who were imaged at many different distances, includes all the same variables as above (field_IRT_stats.csv)
- n/a represent data that were not available
Details for residuals_HR.csv:
- residuals were calculated using the slope of the trend line fit to true heart rate vs IRT-derived heart rate
- n/a represent data that were not available
- Variables:
- animal_ID: ID associated with the individual’s flipper tag
- HR_residuals: residuals of true vs. IRT-derived heart rate
- ambient_temp: air temperature when imaging, in Celsius
- wind_kph: wind speed in kilometers per hour
Details for residuals_RR.csv:
- residuals were calculated using the slope of the trend line fit to true respiration rate vs IRT-derived respiration rate
- n/a represent data that were not available
- Variables:
- animal_ID: ID associated with the individual’s flipper tag
- RR_residuals: residuals of true vs. IRT-derived respiration rate
- ambient_temp: air temperature when imaging, in Celsius
- wind_kph: wind speed in kilometers per hour
Details for wind_HR_success_prob.csv:
- calculates the chance of a successful heart rate measurement at each wind speed (in kilometers per hour)
- n/a represent data that were not available
- Variables:
- wind_kph: wind speed in kilometers per hour, all wind speeds recorded during imaging sessions
- HR_success: 0 for no, not a successful heart rate measurement; 100 for yes, a successful heart rate measurement
- wind_kph.1: wind speed in kilometers per hour, each wind speed is only listed once
- chance of HR success: average of all HR_success for each wind speed
Details for wind_RR_success_prob.csv:
- calculates the chance of a successful respiration rate measurement at each wind speed (in kilometers per hour)
- n/a represent data that were not available
- Variables:
- wind_kph: wind speed in kilometers per hour, all wind speeds recorded during imaging sessions
- RR_success: 0 for no, not a successful respiration rate measurement; 100 for yes, a successful respiration rate measurement
- wind_kph.1: wind speed in kilometers per hour, each wind speed is only listed once
- chance of HR success: average of all HR_success for each wind speed
Details for field_IRT_stats.R:
- all statistics and most figures for the manuscript were created using this code. All required datasets to run the code are included here.
Code/software
The Lambda vue software used to apply Eulerian Video Magnification (EVM) to the infrared videos is available at: https://lambda.qrilab.com/site/
Statistics were done in R 4.4.2 using RStudio version 2024.12.0+467
Access information
This is the only location where these data are publicly available. Contact Caroline Rzucidlo (crzucidlo@whoi.edu / crzucidlo13@gmail.com) with any questions.