Infrared Thermography for Non-Contact Monitoring of Physiological Load in Repetitive Work-like Tasks

Muscle fatigue is a key factor in repetitive physical activities because it progressively reduces force-generating capacity and alters motor control, with consequences for performance, work quality, and musculoskeletal overload risk. In this context, infrared thermography (IRT) is gaining interest as a noninvasive and nonobstructive technique to monitor fatigue-related physiological load by tracking skin temperature changes associated with local metabolism, heat transfer, and crucially perfusion-driven thermoregulatory responses. Unlike approaches focused on mechanical output or neuromuscular activation (e.g., force metrics or sEMG), IRT can provide a spatial map of temperature patterns that are strongly modulated by cutaneous blood perfusion, offering an indirect but informative view of the vascular component of fatigue that is often missed by other commonly used monitoring techniques. This study proposes and tests a replicable IRT protocol focused on a repetitive calf-raise task targeting the gastrocnemius. Five participants performed two conditions: the first involved the test without external load (3 × 1 minute of sessions with 1 minute of rest), while the second was conducted with 0.5 kg. ankle weights (2 × 1 minute of sessions with 1 minute of rest). Thermal data were acquired with a high resolution FLIR system under controlled conditions and processed through thermometric analysis using a static ROI over the central gastrocnemius to extract mean/max/min temperature descriptors. Results showed systematic heating: without load, gastrocnemius maximum temperature increased moderately but consistently; with load, the thermal response was more pronounced and localized, reaching up to ∼ 2.6 °C above baseline in some subjects. Although IRT does not measure fatigue directly, these findings support its sensitivity to local physiological stress and its ability to capture perfusion related thermal dynamics during repetitive work-like actions. Limitations (small sample size, environmental sensitivity, static ROI) are discussed, together with the potential of multimodal integration (e.g., IRT with sEMG and/or NIRS) and interpretive bioheat modelling (Pennes) to improve robustness and physiological interpretability.