生物力学模型导引的心肌运动与材料参数对偶估计

针对传统心脏图像分析方法割裂运动分析和材料分析的局限性和扩展卡尔曼滤波器联合算法引入互协方差矩阵具有收敛慢的问题,提出基于卡尔曼滤波器和扩展卡尔曼滤波器的对偶滤波器算法,该方法用有限元方法表达心脏求解域,连续生物力学模型作为逆问题的约束,在状态空间框架下将运动状态向量和材料参数向量分别置于2组状态空间方程中,用相应的滤波器进行交替迭代估计,从而实现心肌运动状态和材料参数的同时重建.结果表明,该方法能够有效地提高估计精度和收敛速度,缩短计算时间.

Biomechanical model guided dual estimation of myocardial motion and material parameters

The majority of traditional estimation techniques based on medical images deal with cardiac motion analysis and material parameters estimation separately without consideration of the underlying close connection.And in order to improve the slow convergence of Kalman filter(KF) joint estimation method caused by the computation of cross covariance between parameter and state,a dual estimation framework relying on KF/EKF(extended KF) was proposed.A finite element method was employed to represent the computation domain and then a coupled biomechanical-model constrained state-space framework,one for the motion and the other for the material parameter,was formulated.It first generated estimates of heart kinematics with suboptimal material parameter estimates with KF,and then recovered the elasticity property given these kinematic state estimates based on EKF technique.Finally,the coupled filters would estimate motion and material properties simultaneously.The improved accuracy and fast convergence performance of the proposed strategy is demonstrated with synthetic data of varying noises.