A novel approach for segmentation of intravascular ultrasound images

Segmentation of intravascular ultrasound (IVUS) images is a critical step to quantitatively analyze the vascular wall for vascular disease diagnosis and assessment. In this paper, a novel approach is proposed for segmentation of lumen and media-adventitia boundaries from intravascular ultrasound images. The main characteristic of the approach is that different segmentation strategies are utilized respectively for lumen and media-adventitia boundaries according to different IVUS image features. For lumen, the segmentation is carried out by combining the image gradient with fuzzy connectedness model. For media-adventitia boundary, the minimal path based on fast marching model is adopted. The performance of the proposed approach was evaluated over an image database with 180 IVUS image frames of 9 patient cases. The preliminary experimental results show the potential of the proposed IVUS image segmentation approach.