Dynamic biomarkers in ultrasound localization microscopy

Ultrasound localization microscopy (ULM) provides non-invasive, deep-tissue imaging of the microvasculature by tracking millions of individually injected microbubbles, approved for human use, across hundreds of thousands of ultrasound images acquired within minutes. However, state-of-the-art ULM faces significant limitations: it cannot effectively map capillaries and is strongly hindered by tissue motion in the heart. In this work, we present and utilize a novel simulation framework that models connected vascular networks representing the entire vasculature of the mouse brain and human heart to predict the effects of singular value filtering, the mouse skull, and human cardiac motion on dULM image quality. This framework enables the development of novel acquisition sequences and image reconstruction algorithms. Specifically, we demonstrate the feasibility of performing dULM throughout the entire cardiac cycle in rats and pigs in vivo by employing a Lagrangian beamformer that virtually eliminates cardiac motion. Furthermore, we show how tracking microbubbles across thousands of frames allows for the detection of single capillary reporters in the mouse brain through both skin and skull in vivo, providing unprecedented insights into capillary function during neuroinflammation and other pathophysiological conditions.