Philips Research Laboratories, AE Eindhoven, The Netherlands.
Comput Med Imaging Graph. 2012 Apr;36(3):215-26. doi: 10.1016/j.compmedimag.2011.12.005. Epub 2012 Jan 28.
In this paper a method which combines iterative computed tomography reconstruction and coronary centerline extraction technique to obtain motion artifact-free reconstructed images of the coronary arteries are proposed and evaluated. The method relies on motion-vector fields derived from a set of coronary centerlines extracted at multiple cardiac phases within the R-R interval. Hereto, start and end points are provided by the user in one time-frame only. Using an elastic image registration, these points are propagated to all the remaining cardiac phases. Consequently, a multi-phase three-dimensional coronary centerline is determined by applying a semi-automatic minimum cost path based extraction method. Corresponding centerline positions are used to determine the relative motion-vector fields from phase to phase. Finally, dense motion-vector fields are achieved by thin-plate-spline interpolation and used to perform a motion-corrected iterative reconstruction of a selected region of interest. The performance of the method is validated on five patients, showing the improved sharpness of cardiac motion-corrected gated iterative reconstructions compared to the results achieved by a classical gated iterative method. The results are also compared to known manual and fully automatic coronary artery motion estimation methods.
本文提出并评估了一种将迭代重建与冠状动脉中心线提取技术相结合的方法,以获得无运动伪影的冠状动脉重建图像。该方法依赖于从 R-R 间期内多个心动周期中提取的一组冠状动脉中心线得出的运动矢量场。为此,仅在一个时相中由用户提供起始点和结束点。使用弹性图像配准,这些点会传播到所有剩余的心动周期。因此,通过应用基于半自动最小成本路径的提取方法来确定多相位三维冠状动脉中心线。相应的中心线位置用于确定从一相位到另一相位的相对运动矢量场。最后,通过薄板样条插值获得密集的运动矢量场,并用于对选定的感兴趣区域进行运动校正迭代重建。该方法在五名患者身上进行了验证,结果表明与经典门控迭代方法相比,心脏运动校正门控迭代重建的清晰度得到了提高。结果还与已知的手动和全自动冠状动脉运动估计方法进行了比较。