Analysis of Breathing Patterns from Thermal Images Using an Automated Segmentation Method

Breathing is one of the important vital signs in diagnosing and monitoring for patients’ treatment and disease. Few modalities have been used to evaluate breathing activity such as respiratory belt, thermistor and capacitive sensor. However, these requires external attachments such as electrode or sensor which might be inconvenience over long period of time. Hence, we proposed the use of thermography as a contactless monitoring device. In this study, inspiration time and expiration time of three different breathing patterns such as normal, prolonged and rapid breathing patterns were measured by using the thermography. Thermal images obtained from the subjects were processed and analysed by using an automated segmentation method which integrate the knowledge of edge-based and region-based segmentation methods into the algorithm developed. The algorithm developed in this study has shown that the tracker was able to segment the region of interest of the thermal images automatically and it provides a more accurate and stable results than manual calculation method. Thus, three different types of breathing patterns could be identified based on the inspiration time to expiration time ratio. Results shows that there was less than 5% of relative error which suggest the benefit of this algorithm.