Corina Barbalata and Leonardo S. Mattos

Abstract-This research proposes a new system for the localization, segmentation and classification of lesions in the larynx based on endoscopic narrow-band imaging video. The developed algorithm starts by detecting the approximate location of the lesion in the vocal cords, which is followed by the segmentation of the blood vessels present in that region. The algorithm continues with the measurement of geometric features of the blood vessels map, based on which an statistical analysis provides the probability of malignancy. To our best knowledge, this is the first system developed for the segmentation and analysis of the blood vessels in the vocal cords.