Automatic pediatric otitis detection by classification of global image features

This paper proposes a preliminary study regarding the implementation of a pediatric otorhinolaryngology diagnosis support system for the visual diagnosis of the eardrum, based on the automatic evaluation by digital image processing techniques of the otoscopic digital color images of the eardrum. We compare different color descriptors and classification algorithms (k-Nearest Neighbor, Decision Trees, Linear Discriminant Analysis, Naïve Bayes, Multi Layer Neural Networks and Support Vector Machines) and try to choose the best solution in terms of performance/computation to be included in an integrated framework.