MYOCARDIAL FUNCT USING SPATIOTEMPORAL FINITE ELEMENT MESH MODEL & VELOCITY DATA

点击翻译按钮获取中文摘要

Introduction: Phase contrast velocity data can be used to compute motion and deformation of a myocardial region. We present a method that uses a global spatiotemporal finite element model to improve the reproducibility of the computation while retain high accuracy, as compared to a local computation method. Methods and Results: We consider a partition of a large myocardial region with non-overlappingtime-varying basic mesh elements. Nodal points of the mesh covering the region are selected at a reference state. These material points and corresponding spatiotemporal trajectories define the time-varying domain of each mesh element. Within each element, we model the kinematics using finite element shape functions. In implementation, with a reference state mesh configuration defined by the user, and initial estimates of the subsequent mesh configurations obtained by independent tracking of node points, we iteratively refine the mesh configuration estimates by fitting the velocity data to the spatiotemporal finite element model. Reproducibility of the computed quantities can be analyzed after after each computation. In a validation study, we synthesized a 2-D nonuniform deformation data set with an apprent SNR of 45. We computed the motion and deformation quantities from analyzing a single triangular mesh element in case 1 and analyzing an extended mesh configuration that consisted of the same element plus three additional neighboring elements of compariable size in case 2. Repeating the data synthesis and analysis 1000 times for mean and standard deviation measurements showed that the averages of the estimated quantities were in excellent agreement with the true values in both cases. Comparing the results for the same mesh element in the two cases indicated that the extended mesh modeling resulted in lower standard deviation but comparably high accuracy. Summary and Conclusion: Analyzing a single element domain with the present method and the previous local method have about the same reproducibility. With the present method, extending the finite element mesh by adding more triangular elements achieves increasing reproducibility. At the same time, due to the piecewise parameterization, high estimation accuracy, comparable to that of the local method, can still be obtained.