Automated segmentation and temperature extraction from thermal images of human hands, shins and feet

Abstract

The use of thermography has been considered for a wide range of medical applications, which often require the extraction of temperature values from specific points of interest on the human body. However, temperature extraction is typically carried out manually, rendering the process both lengthy as well as highly subjective. In this work we propose a number of methods to automatically segment and extract temperature readings from thermal images of human hands, shins and feet that can be employed in several clinical applications. Tests conducted using thermal images from the body regions of interest have shown that the implemented feature detection and region growing methods can provide accurate results with a correct detection rates of 100% in the case of the shin regions and detection rates above 90% for hand and foot regions.