Coronary cine-angiography segmentation incorporating low-rank priori

In this paper, a new approach of segmenting cine coronary angiography is proposed. Rather than treating each frame of a video sequence individually and applying the segmentation on them separately, our approach tackles the segmentation of all the frames as a whole via incorporating the prior knowledge that the background components in all frames have much strong correlation mutually than vessels do, so our approach is global in the time dimension. We have applied this approach to the segmentation of clinical data and obtain encouraging result.