Oriented Gaussian mixture models for nonrigid 2D/3D coronary artery registration

Area of Science:

• Medical Imaging
• Computer-Aided Surgery
• Biomedical Engineering

Background:

• Accurate 2D/3D registration of patient vasculature is crucial for guiding percutaneous coronary interventions.
• Aligning pre-interventional computed tomography angiography (CTA) with interventional X-ray angiography aids procedural navigation.

Purpose of the Study:

• To present a novel feature-based 2D/3D registration framework using probabilistic point correspondences.
• To demonstrate its effectiveness in aligning 3D coronary artery centerlines from CTA with their 2D projections from X-ray angiography.

Main Methods:

• The framework extends Gaussian Mixture Model (GMM) based point-set registration to the 2D/3D setting with a modified distance metric.
• Incorporation of orientation information into the GMM registration process.
• Utilized a statistical shape model for nonrigid vessel centerline registration.

Main Results:

• The oriented GMM registration achieved a median accuracy of 1.06 mm.
• Demonstrated a convergence rate of 81% for nonrigid vessel centerline registration across 12 patient datasets.
• Outperformed iterative closest point, GMM without orientation, and two other published methods.

Conclusions:

• The proposed oriented GMM registration framework offers superior accuracy and robustness for 2D/3D coronary artery registration.
• This method enhances guidance for percutaneous coronary interventions by improving the alignment of pre-procedural and intra-procedural imaging data.