A quantitative evaluation of the three dimensional reconstruction of patients' coronary arteries

Area of Science:

• Medical Imaging
• Computational Anatomy
• Cardiovascular Science

Background:

• Angiographers mentally reconstruct 3D coronary artery structures from 2D views.
• Computer graphics can aid in visualizing coronary 3D structure and predicting optimal angiogram views.

Purpose of the Study:

• Evaluate a new computer method for 3D reconstruction and visualization of human coronary arteries.
• Assess the accuracy and reproducibility of the 3D reconstruction technique.

Main Methods:

• Reconstructed biplane left coronary angiograms from 30 patients using the new computer method.
• Assessed accuracy by comparing actual vessel centerlines to projected 3D model centerlines and comparing models from different angiogram pairs.
• Evaluated inter- and intraobserver variability of reconstructions.

Main Results:

• Average displacement between actual and reconstructed vessel centerlines was 1.64 mm.
• Average displacement between 3D models from different views was 7.08 mm.
• Intraobserver variability was 5.22 mm; interobserver variability was 6.6 mm.

Conclusions:

• The 3D reconstruction algorithm is mathematically accurate and reproducible.
• Measurement errors are clinically acceptable, providing an accurate representation of angiogram information.
• The technique is clinically useful for displaying complex coronary arterial structures in 3D.