Adaboost to classify plaque appearance in ivus images

Abstract

Intravascular Ultrasound images represent a unique tool to analyze the morphological vessel structures and make decisions about plaque presence. Texture analysis is a robust way to detect and characterize different kind of vessel plaques. In this article, we make exhaustive comparison between different feature spaces to optimally describe plaque appearance and show that applying advanced classification techniques based on multiple classifiers (adaboost) significantly improves the final results. The validation tests on different kind of plaques are very encouraging. © Springer-Verlag 2004.