Continuous respiration monitoring is an important tool in assessing the patient’s health and diagnosing pulmonary, cardiovascular, and sleep-related breathing disorders. Various techniques and devices, both contact and contactless, can be used to monitor respiration. Each of these techniques can provide different types of information with varying accuracy. Thermal cameras have become a focal point in research due to their contactless nature, affordability, and the type of data they provide, i.e., information on respiration motion and respiration flow. Several studies have demonstrated the feasibility of this technology and developed robust algorithms to extract important information from thermal camera videos. This paper describes the current state-of-the-art in respiration monitoring using thermal cameras, dividing the system into acquiring data, defining and tracking the region of interest, and extracting the breathing signal and respiration rate. The approaches taken to address the various challenges, the limitations of these methods, and possible applications are discussed.

Thermal imaging is a non-contact method for monitoring respiration activity during sleep. In this study, we evaluated its clinical application during overnight recordings in a sleep clinic. Five thermal cameras were used to detect breaths, the estimated respiration rate (RR), and inter-breath intervals (IBIs) in seven adults undergoing diagnostic polysomnography (PSG). Forty-five minutes of recordings were selected, consisting of 12 motionless and event-free segments. The thermal videos were processed using an adapted pre-existing thermal video processing algorithm. The respiration signals generated with the thermal cameras were validated against simultaneously recorded signals from the PSG system, the current gold standard for monitoring sleep. The results show a mean absolute error (MAE) ranging between 0.64 and 0.91 breaths per minute for the RR. Breath detection showed a sensitivity of 96.3%, and a precision of 94.1%. The MAE obtained between IBIs was 0.48 s, and the mean IBI variability difference recorded was 3.9 percentage points. In addition, the results from this clinical study show that the use of all five cameras and a single camera revealed no statistically significant differences, demonstrating the work towards a robust system. This first study of thermal cameras for the assessment of respiration in a clinical setting shows us the potential application of thermal imaging in clinical practice for respiration monitoring and establishes a foundation for further implementation in assessing sleep-disordered breathing.