Sliding Window Beamforming Enhances Speckle Resolution in High Frame Rate Imaging

Speckle resolution determines the accuracy and precision of ultrasound-based displacement estimation. At high frame rates, time constraints lead to under-sampling of the field of view and result in a suboptimal point spread function, especially in the lateral dimension. Minimum Variance beamforming (MV) is theorized to improve lateral resolution, but several in vivo studies have shown little or no improvement. We hypothesize that this can be overcome by applying MV to overlapping 2D windows of the imaging grid, calculating the apodization weights to optimize each window, rather than applying the algorithm to one entire line of depth per iteration. This sliding window approach to minimum variance (SWMV) allows the beamformer to adapt to variations in noise levels as a function of width as well as depth in order to more effectively suppress off-axis interferences. SWMV is shown to outperform delay-and-sum beamforming as well as conventional MV in vivo. In B-mode images of the myocardium, the average radial speckle size was 4.3 pixels with DAS and 4.1 pixels with conventional MV, while SW yielded a speckle size of 2.6 pixel at FWHM of the autocorrelation function. The observed improvements were consistent across different subjects and different transmission sequences including coherently compounded diverging wave imaging of the myocardium, and coherently compounded plane wave imaging of the carotid artery. In the carotid artery, speckle resolution was also greatly improved, and in the case of a stroke patient, the imaged plaque was enhanced both in terms of resolution and contrast against the vessel lumen. Cardiac systolic displacement estimation was also performed and shows superior tracking with the proposed beamformer; manual segmentation was used to quantify the accuracy of displacement estimation, and the SW beamformer results in an improved IoU of 72% compared to 61% with conventional DAS beamforming.