Ultrasound Localization Microscopy with reduced frame rates and short acquisition times using bidirectional 2-D interpolation

Ultrasound Localization Microscopy (ULM) enables high-resolution imaging of microvasculature but faces significant challenges due to its dependency on high frame rates (in kHz) and long acquisition times, both of which limit its clinical applicability. To address these challenges, we propose a novel bidirectional Radial Basis Function (RBF)-based interpolation technique that reconstructs missing data by leveraging spatiotemporal information from orthogonal x-t and z-t planes. Results from in silico and in vivo datasets, including rat brain and kidney scans, show the approach preserves microvascular depiction and resolves velocity maps. Specifically, the technique enables a tenfold reduction in frame rate (from 1 kHz to 100 Hz) and a 3- to 4-fold compression of acquisition time without significantly compromising super-resolution image quality. Fourier Ring Correlation (FRC) and Dice similarity validate the effectiveness of the proposed technique. From 100-Hz frame rate, the proposed technique achieves a FRC and Dice score (for Density maps) of 18.9 μm and 0.82, respectively. For comparison, images generated at 100 Hz (before interpolation) achieve FRC and Dices score of 22 μm and 0.19, respectively (FRC at 1 kHz: 11.5 μm). By enabling low-frame-rate and short-acquisition ULM, our technique significantly enhances ULM’s clinical feasibility. Future work will integrate motion correction to further optimize dynamic profiles.