Improving Harmonic Motion Estimation with Phase-Based Estimators for Magnetomotive Ultrasound Imaging

Magnetomotive Ultrasound Imaging relies on estimates of the harmonic motion in tissue from ultrasound data. It was shown that data containing harmonic motion are different in spectral content from those containing uniform motion as in classical Doppler methods. However, common computationally efficient phase-based motion estimators are similar to Doppler processing. Simulation data indicate that even in case of noise-free data, the phase-based estimators exhibit a significant error in a typical magnetomotive ultrasound scenario. An improvement of the estimator, measured as a reduction of over 50 % in mean absolute error, was achieved in the simulation scenario by prefiltering radio frequency data components not associated with motion and reducing the bandwidth of radio frequency data or by averaging motion estimates obtained in different frequency bands. Thereby, it was shown that a substantial improvement of the estimators is possible without increasing the computational effort significantly. However, it was also pointed out that further improvement in case of small harmonic displacements ratio seems possible only with different estimators derived from more sophisticated signal models.