Semi-Automatic Reconstruction of Patient-Specific Stented Coronaries based on Data Assimilation and Computer Aided Design.

Abstract

The interplay between geometry and hemodynamics is a significant factor in the development of cardiovascular diseases. This is particularly true for stented coronary arteries. To elucidate this factor, an accurate patient-specific analysis requires the reconstruction of the geometry following the stent deployment for a computational fluid dynamics (CFD) investigation. The image-based reconstruction is troublesome for the different possible positions of the stent struts in the lumen and the coronary wall. However, the accurate inclusion of the stent footprint in the hemodynamic analysis is critical for detecting abnormal stress conditions and flow disturbances, particularly for thick struts like in bioresorbable scaffolds. Here, we present a novel reconstruction methodology that relies on Data Assimilation and Computer Aided Design. The combination of the geometrical model of the undeployed stent and image-based data assimilated by a variational approach allows the highly automated reconstruction of the skeleton of the stent. A novel approach based on computational mechanics defines the map between the intravascular frame of reference (called L-view) and the 3D geometry retrieved from angiographies. Finally, the volumetric expansion of the stent skeleton needs to be self-intersection free for the successive CFD studies; this is obtained by using implicit representations based on the definition of Nef-polyhedra. We assessed our approach on a vessel phantom, with less than 10% difference (properly measured) vs. a customized manual (and longer) procedure previously published, yet with a significant higher level of automation and a shorter turnaround time. Computational hemodynamics results were even closer. We tested the approach on two patient-specific cases as well. The method presented here has a high level of automation and excellent accuracy performances, so it can be used for larger studies involving patient-specific geometries.