Ultrasound localization microscopy (ULM) is a super-resolution imaging technique that exploits intravascular contrast microbubbles (MBs) to break the diffraction limit. Conventional MB localization strategies require spatially sparse MB distributions, leading to long data acquisition times. Several strategies to localize MBs at higher concentrations have been proposed in the literature, such as spatiotemporal filtering for data splitting, sparse recovery, multi-feature localization/tracking, and deep-learning. For small animal imaging that does not require deep imaging penetration, a potential strategy is increased imaging frequency, which reduces the spatial extent of the MB point-spread function (PSF) and enables isolated MB features at higher concentrations. Here we investigate ULM performance using three different imaging frequencies (18, 28, and 40 MHz), for the same mouse brain in a single imaging session. All other imaging parameters and experimental parameters were kept as consistent as possible. Our results indicate that higher imaging frequency resulted in spatially smaller MB PSFs, which increased the proportion of MB localizations suitable for ULM tracking. The fidelity of ULM imaging was improved with increasing frequency, as evidenced by less noisy vascular reconstruction, which can be attributed to smaller PSFs and thinner elevational beam width. We demonstrate a j-shaped MB trajectory that transitions from arteriole to venule in the cortex, which implies capillary-level vascular flow.
Pengfei Song (2025) studied this question.