Segmentation Algorithms for Thermal Images

Biomedical techniques and applications are being developed and placed at the service of clinicians. An example is medical thermography, which is being used more often in the detection of certain diseases and also in pain distribution. Current thermography processing software has some limitations mainly because it is developed for general applications and does not allow the identification of a Region Of Interest (ROI) with a specific anatomic shape. Current commercial software usually uses regular prismatic shapes for the definition of these regions, such as, rectangles, squares, circles and/or ellipse that poorly define complex geometric regions. These shapes present limitations when they do not fit with the complex geometric shape of the area that is to be characterized, either by the exclusion or the inclusion of irrelevant data in the evaluation of the thermal images. This particular limitation is observed no matter how accurate the definition of the ROI is. In order to improve characterization of thermal images, a computational application was developed. The limitations of existing software applications was overcome by designing an application that allows choosing any ROI, independently of its geometric shape and optimize it for further processing. This research work presents several segmentation algorithms and a comparison of untreated and optimized ROI's.