A comparison of intelligent classifiers of thermal patterns in diabetic foot

The Diabetic foot ulcer is one of the major complications of Diabetes Mellitus. A suitable technique in detection and prevention of this complication is the analysis of temperature distribution in the plantar region by means of infrared thermography, a non-invasive and non-contact technique. However, due to some complications either from human interactions or related to the data, an emergent alternative is the use of computer-aided diagnosis, taking part in the support decision, and analysis systems with high accuracy rates of classification. This work addresses the employment of computational intelligence-based methods in the classification of thermal images, aiming to compare two of the most powerful techniques such as Support Vector Machines and the Multilayer Perceptron. Both techniques are compared with the recent and renewed usage of Deep Learning structures, a massive structure of artificial neural networks. The research goes from the automatic extraction of the region of interest (ROI) by means of evolutionary algorithms until the final comparison of these three relevant classifiers. The high accuracy of the obtained results shows the utility of these methods as auxiliary tools in the medical diagnosis.