Infrared Thermography for Measuring Elevated Body Temperature: Clinical Accuracy, Calibration, and Evaluation

Area of Science:

• Medical Imaging and Diagnostics
• Biomedical Engineering
• Public Health Surveillance

Background:

• Infrared thermographs (IRTs) show potential for detecting elevated body temperatures (EBT), crucial for clinical settings and epidemic control.
• Optimal calibration methods for IRTs and their comparative clinical performance against non-contact infrared thermometers (NCITs) remain unclear.

Purpose of the Study:

• To compare different methods for IRT calibration and establish best practices for assessing IRT performance in detecting EBT.
• To compare the clinical accuracy of IRTs with NCITs.

Main Methods:

• A clinical thermographic imaging study involving over 1000 subjects.
• Acquisition of facial temperature data and reference oral temperatures for calibrating two IRT systems using seven regression methods.
• Evaluation of IRT clinical accuracy using metrics like clinical bias, repeatability, root-mean-square difference, and sensitivity/specificity.

Main Results:

• A constant offset calibration approach demonstrated better EBT detection capabilities.
• Inner canthi or full-face maximum temperatures yielded the highest clinical accuracy, achieving a clinical bias (Δcb) between ±0.03 °C and standard deviation (σΔcb) < 0.3 °C.
• Optimal calibration resulted in sensitivity/specificity between 84% and 94%; forehead-center measurements showed reduced performance for both IRTs and NCITs.

Conclusions:

• Novel insights into methods and metrics for assessing the clinical accuracy of IRTs are provided.
• Calibration methods optimizing accuracy in the 37-38.5 °C range are most effective for measuring EBT.
• Optimally calibrated IRTs can offer superior performance compared to NCITs for detecting elevated body temperatures.