Lacrimal caruncula thermography overestimates rectal temperature in horses: Influence of camera type and measurement distance.

Overview

• This study investigated whether measuring the temperature of the lacrimal caruncula (a small area near a horse’s eye) with infrared thermography can accurately estimate rectal temperature, the gold standard for body temperature in horses.
• The research compared two types of infrared cameras and examined how the distance between the camera and horse affected temperature accuracy.

Background and Motivation

• Rectal temperature (RT) measurement is considered the most reliable method to assess body temperature in horses but involves physical handling, which may affect welfare and practical usability.
• Infrared thermography of the lacrimal caruncula (LC) offers a noncontact way to estimate temperature but may be less accurate depending on device and method.
• There was a need to systematically evaluate the agreement between LC thermography and rectal temperature and the impact of factors like camera type and measurement distance.

Study Objectives

• Assess agreement between LC thermography and rectal temperature in horses.
• Compare two infrared thermal cameras (IR FlexCam S and UNI-T UTi165A) for accuracy.
• Determine how camera-to-subject distance (0.5, 1, and 2 meters) influences measurement bias and variability.

Methods

• Participants: 84 healthy Thoroughbred horses were used, with three temperature measurements taken per camera per horse.
• Measurements: LC temperature recorded at three distances (0.5 m, 1 m, 2 m) for both cameras.
• Analysis:
  • Agreement between LC temperatures and rectal temperatures was assessed using Bland-Altman analysis to identify mean bias (difference) and limits of agreement (LoA).
  • Linear regression was performed to detect proportional bias by analyzing the relationship between measurement differences and mean temperature.

Key Findings

• Both cameras consistently overestimated temperature at the lacrimal caruncula compared to rectal temperature.
• Bias increased with greater camera distance from the subject:
  • IR FlexCam S biases: 3.75°C (0.5 m), 3.71°C (1 m), 5.28°C (2 m).
  • UNI-T UTi165A biases: 2.97°C (0.5 m), 3.85°C (1 m), 5.21°C (2 m).
• Limits of agreement widened as distance increased, indicating greater variability and less precision at farther distances.
• Negative regression slopes indicated proportional bias for both devices, meaning the degree of overestimation varied depending on temperature levels.
• The UTi165A camera demonstrated narrower limits of agreement and better repeatability at the shortest distance (0.5 m) compared to the FlexCam.

Conclusions and Implications

• LC thermography overestimates the rectal temperature in horses, and this overestimation worsens as the camera moves farther from the subject.
• Although the UTi165A showed better accuracy and consistency at close range, neither camera achieved agreement sufficient to replace rectal temperature measurements.
• For LC thermography to be useful as a practical screening tool in equine veterinary practice, careful selection of the thermal camera, short measurement distances, and standardized measurement protocols are necessary.
• Limitations:
  • Study was conducted indoors, limiting real-world environmental variation assessment.
  • Only one anatomical site (lacrimal caruncula) was used for thermography measurements.
  • Only healthy Thoroughbred horses were included, which limits generalizability to other breeds or horses with health issues.
  • Only two thermal camera models were tested, so findings may not extrapolate to other devices.

Overall Significance

• This research highlights the challenges and limitations of using infrared thermography at the lacrimal caruncula as a noninvasive proxy for core body temperature in horses.
• It provides valuable guidance on the impact of device choice and measurement distance, emphasizing the need for standardization to improve accuracy and repeatability in clinical or field settings.
• Further research could explore other anatomical sites, additional camera models, and include horses with various health statuses to improve robustness and applicability.