An Automated Lumen Contour Segmentation In Intravascular Ultrasound Images Using An Efficient Variational Approach

Intravascular ultrasound (IVUS) is a minimally invasive medical imaging that produces cross-sectional images of blood vessels. IVUS processing could help experts to put in accurate diagnosis and proper treatment by measuring stenosis size, and the disposal location. Lumen border detection in IVUS images represents a necessary step just before the reliable quantitative valuation of atherosclerosis. In this work an automated technique is proposed for the lumen border detection in IVUS images. At first a novel variational minimax algorithm is used to make local threshold. Then simple classifier classified threshold to extract lumen border. The method is validated through a comparison with expert manual segmentation on a challenging database and compared with recent methods for the lumen border detection in IVUS. The obtained averaged values for the mean Hausdorff distance and the percentage of Jaccard similarity index are 0.361mm and 95.0%, respectively.