Iterative Closest Curve: A Framework for Curvilinear Structure Registration Application to 2D/3D Coronary Arteries Registration

Abstract

Abstract. Treatment coronary arteries endovascular involves catheter navigation through patient vasculature. The projective angiography guidance is limited in the case of chronic total occlusion where occluded vessel can not be seen. Integrating standard preoperative CT angiography information with live fluoroscopic images addresses this limitation but requires alignment of both modalities. This article proposes a structure-based registration method that intrinsically preserves both the geometrical and topological coherencies of the vascular centrelines to be registered, by the means of a dedicated curveto-curve distance pairs of closest curves are identified, while pairing their points. Preliminary experiments demonstrate that the proposed approach performs better than the standard Iterative Closest Point method giving a wider attraction basin and improved accuracy.

A tree-topology preserving pairing for 3D/2D registration

“…into the Iterative Closest Curve (ICC)-framework introduced in [2]. It aims at solving the registration problem of (1) where the distance D is based upon curve-pairings instead of point-pairings.…”

“…We propose to consider all paths p extracted from the graph X such that p[1] = b and p[#] belongs to N (P (C[#]), R C ). Instead of only computing the shortest path in the 2D graph as in [2], we found more realistic to look for all 2D paths having a length that is compatible with the projection of the 3D curve. Thus, the length difference |L (P (C)) − L(p)| should be smaller than an expected variation ∆L C .…”

“…Instead of penalizing incoherences in pairing between points along the same vessel, both [2,4] consider the matching of entire vessels extracted from the main bifurcation of the 3D tree to the segmentation extremities. While [4] shows the feasibility of such an approach in clinical practice by manually providing 3D/2D vessel correspondences and segmentation, [2] proposes a framework that extends the ICP principle to curves and introduce the Iterative Closest Curve (ICC) algorithm. In addition of the general framework that provides a way to solve the problem of registering curvilinear structures, they propose a curve-pairing procedure that matches vessels starting from the main bifurcation to the segmentation extremities.…”

“…Instead of ignoring centerlines as in the matching process described in [20], that is only based on bifurcation matching, the pairing pro-cedure described in this article matches tree centerlines while ensuring connectivity at bifurcation points. We integrate the proposed pairing procedure into the framework designed to register curves, introduced in [2], by replacing their pairing procedure by our tree-preserving one. Contrary to [2] that considered the pairing of proximal vessels multiple times without any control on the coherence, the proposed approach ensures a oneto-one matching of curves (and points) in the model while preserving the tree topology.…”

“…We integrate the proposed pairing procedure into the framework designed to register curves, introduced in [2], by replacing their pairing procedure by our tree-preserving one. Contrary to [2] that considered the pairing of proximal vessels multiple times without any control on the coherence, the proposed approach ensures a oneto-one matching of curves (and points) in the model while preserving the tree topology. By imposing such consistency, we expect the resulting pairing to be more relevant regarding the clinical application.…”