On 3D Cardia-Cerebrovascular Segmentation

Cardia-cerebrovascular disease is the most common diseases threatening the human health seriously,and it is at the head of causing human death.Segmentation is a key step toward an accurate three dimensional visualization which is significantly helpful for clinical diagnosis and treatment of related diseases.In view of the location particularity,the shape complexity,and the intensity variability of the cardio-cerebral vessels,although there is a tremendous amount of past and on-going dedicated research,its segmentation still remains a challenging task.In this paper,authors present a technical review on state-of-the-art literature for 3D cardia-cerebrovascular segmentation from MRA and CTA images.Firstly,four types of segmentation techniques,including pattern recognition approaches,tubular structure based approaches,centerline-based approaches,active contour model approaches,are discussed.Secondly,for each technique,The comments,on their performance,as well as scope and some aspects to be improved,are presented.Finally,developing direction and the improving strategies are concluded and identified.