Centerline constrained minimal path propagation for vessel extraction

Vessel segmentation takes a key step toward accurate visualization, diagnosis and quantification of vascular pathologies. This paper proposes a Centerline Constrained Minimal Path Propagation (CCMPP) algorithm for vascular extraction. The proposed CCMPP method includes two minimal path propagations. In the first minimal path propagation, a novel back-tracing accumulation strategy is used to extract vessel centerlines, then the centerline information serves as a constraint for the vessel extraction in the second minimal path propagation. For the proposed approach, only one start point is needed to be set, and no input of start and end points is required for the whole centerline and vessel extraction. We validate the utility of the proposed approach by using 3-D angiography data.