Segmentation of artery wall in coronary IVUS images: aprobabilistic approach

摘要

Intravascular ultrasound (IVUS) images represent a unique tool toanalyze the morphology of arteries and vessels (plaques, etc). The poorquality of these images makes traditional segmentation algorithms (suchas edge detection) fail to achieve the expected results. We present aprobabilistic flexible template to separate different regions in theimage. In particular, we use elliptic templates to model and detect theshape of the vessel inner wall in IVUS images. The use of elliptictemplates forces a global probabilistic approach, that makes use ofimage statistics inside regions. We present the results of successfulsegmentation obtained from 12 patients undergoing stent treatment. Aphysician team has validated these results