Sub-micron Tissue Motion Estimation for High Framerate Ultrasound using Complex Gradient Optimization

This work introduces a complex-valued gradient optimization approach for precise motion estimation and correction in high-framerate ultrasound imaging. The method is hypothesized to outperform non-optimization approaches such as transverse oscillation, speckle tracking, and an intensity-based optimization approach. This is then hypothesized to result in quantifiable image quality improvements when used for motion correction during microvascular flow imaging with ultrasound. Motion estimation is achieved by iteratively aligning neighboring beamformed ultrasound frames to minimize the per-pixel magnitude of the complex-valued difference. In ultrasound microvascular simulations with a maximum global tissue motion of 6.6 mm/s and a maximum displacement of 1.7 mm, the mean motion error compared to prior phase-based techniques is reduced by a factor of 5.1 for lateral and 2.4 for axial motion. Comparing neighbor-frame complex gradient optimization-based motion estimation to transverse oscillation motion estimation, median lateral and axial motion error is, respectively, reduced to 0.9 and 0.17 µm from an original 5.08 and 0.56 µ m, demonstrating sub-micrometer accuracy motion estimation with a 154 µm transmit wavelength. Tested on in vivo ultrasound data of a Sprague-Dawley rat kidney, the peak residual tissue signal after echo cancellation is reduced by approximately 11 dB compared to prior techniques. In summary, motion estimation using complex gradient optimization has sub-micron accuracy in microvascular simulations and is shown to reduce residual tissue signal for in vivo ultrasound microvascular imaging.