A robust bidirectional motion-compensated interpolation algorithm to enhance temporal resolution of 3D echocardiography

Abstract

High frame rate three-dimensional echocardiography (3DE) improves the assessment of the cardiac dynamics by providing a realistic view of the cardiac structures. However, since the field of view in 3D is large and the speed of sound in tissue is limited, one must sacrifice some spatial resolution to achieve a temporally smoothed sequence of 3DE data. Motion-compensated frame interpolation provides a means to enhance perceived temporal resolution without degrading the spatial resolution. Among the existing interpolation techniques, motion-compensated interpolation based on 3D speckle tracking (3DST) is very simple to implement and can be applied in real time since it is well suited for parallel processing. However, the performance of this approach deteriorates in low frame rate 3DE due to large inter-frame displacements. In this paper, we propose a new 3DST-based motion-compensated interpolation algorithm based on a multiscale bidirectional motion estimator. The proposed estimator improves the reliability of the motion estimate at the coarsest scale by utilizing a multidirectional speckle tracker. Moreover, the proposed 3DST algorithm requires smaller search regions for an equivalent search effect compared to the existing algorithms; therefore, it reduces the computation time of 3DST. To reduce the interpolation artifacts caused by decorrelation of the speckle pattern and inaccurate motion estimates, we develop an adaptive interpolation scheme based on robust statistics and the structural similarity index. Experimental results demonstrate that the proposed algorithm, compared to an existing 3DST-based algorithm, greatly reduces the interpolation artifacts such as ghosting and blurring, especially near fast-moving cardiac structures.