Early Sepsis Recognition Based on Ear Localization using Infrared Thermography

Systematic screening is crucial for early diagnosis of sepsis. Detecting abnormal body temperature patterns can accurately predict sepsis before any other symptoms of infection. Therefore, we suggested using thermography as a non-invasive tool capable of continuously measuring body temperature patterns and detect abnormalities. One such pattern is the temperature difference (ETD) between body extremities such as outer ear or tip of the nose, and the core temperature such as the inside of the ear or the eye. Specifically, in sepsis, the ETD decreases from about 5-6F to about 1F. Thus, we propose a fully automated methodology for calculating the core vs. extremity temperature difference based on ear localization using thermal images and the fuzzy C-means (FCM) clustering algorithm. On a pilot dataset that consisted of 3674 face images from three human subjects, we achieved a 96.2% accuracy for ear localization. Consequently, we were able to accurately detect temperature differences between 1 and 6 F. The proposed methodology may be used in intensive care units (ICU) for early sepsis recognition.