Klein J L, Hoff J G, Peifer J W, Folks R, Cooke C D, King S B, Garcia E V
Emory University School of Medicine, Atlanta, GA, USA.
Int J Card Imaging. 1998 Apr;14(2):75-87. doi: 10.1023/a:1005903705300.
Through extensive training and experience angiographers learn to mentally reconstruct the three dimensional (3D) relationships of the coronary arterial branches. Graphic computer technology can assist angiographers to more quickly visualize the coronary 3D structure from limited initial views and then help to determine additional helpful views by predicting subsequent angiograms before they are obtained.
A new computer method for facilitating 3D reconstruction and visualization of human coronary arteries was evaluated by reconstructing biplane left coronary angiograms from 30 patients. The accuracy of the reconstruction was assessed in two ways: 1) by comparing the vessel's centerlines of the actual angiograms with the centerlines of a 2D projection of the 3D model projected into the exact angle of the actual angiogram; and 2) by comparing two 3D models generated by different simultaneous pairs on angiograms. The inter- and intraobserver variability of reconstruction were evaluated by mathematically comparing the 3D model centerlines of repeated reconstructions.
The average absolute corrected displacement of 14,662 vessel centerline points in 2D from 30 patients was 1.64 +/- 2.26 mm. The average corrected absolute displacement of 3D models generated from different biplane pairs was 7.08 +/- 3.21 mm. The intraobserver variability of absolute 3D corrected displacement was 5.22 +/- 3.39 mm. The interobserver variability was 6.6 +/- 3.1 mm.
The centerline analyses show that the reconstruction algorithm is mathematically accurate and reproducible. The figures presented in this report put these measurement errors into clinical perspective showing that they yield an accurate representation of the clinically relevant information seen on the actual angiograms. These data show that this technique can be clinically useful by accurately displaying in three dimensions the complex relationships of the branches of the coronary arterial tree.
通过广泛的训练和经验积累，血管造影师学会在脑海中重建冠状动脉分支的三维（3D）关系。图形计算机技术可以帮助血管造影师从有限的初始视图中更快地可视化冠状动脉的3D结构，然后通过在获取后续血管造影之前预测其图像，帮助确定更多有用的视图。
通过重建30例患者的双平面左冠状动脉造影，评估一种促进人类冠状动脉3D重建和可视化的新计算机方法。从两个方面评估重建的准确性：1）将实际血管造影的血管中心线与3D模型投影到实际血管造影精确角度的二维投影的中心线进行比较；2）比较由不同同步血管造影对生成的两个3D模型。通过数学比较重复重建的3D模型中心线，评估观察者间和观察者内重建的变异性。
30例患者的14662个血管中心线点在二维中的平均绝对校正位移为1.64±2.26mm。由不同双平面组生成的3D模型的平均校正绝对位移为7.08±3.21mm。观察者内绝对3D校正位移的变异性为5.22±3.39mm。观察者间变异性为6.6±3.1mm。
中心线分析表明，重建算法在数学上是准确且可重复的。本报告中的数据将这些测量误差置于临床背景下，表明它们能够准确呈现实际血管造影中所见的临床相关信息。这些数据表明，该技术通过在三维中准确显示冠状动脉树分支的复杂关系，在临床上可能具有实用性。
