Cerebrovascular Structure-Flow Gradients Revealed by Volumetric Ultrasound Localization Microscopy

Abstract

Characterizing the joint organization of microscale cerebrovascular structure and flow in vivo is challenging. Volumetric ultrasound localization microscopy (ULM) provides spatially resolved measures of microvascular structure and flow from the same acquisition, but how these features are jointly organized across brain networks remains unclear. Here, we characterized local vascular phenotypes from transcranial volumetric ULM in mice and examined whether they contained reproducible low-dimensional organization. Local measures of vascular structure and flow dynamics were summarized within corresponding brain regions across mice. Subject-specific vascular phenotype similarity matrices were combined into a group-consensus graph and analyzed using diffusion embedding. Two leading gradients, showing distinct spatial distributions across brain networks, were reproducibly recovered across subject and spatial scales. The principal gradient was associated with local vascular density and flow complexity, defining a vascular structure-related axis. The secondary gradient was associated with flow speed and acceleration, defining a flow-dynamics-related axis. These gradients provide a low-dimensional representation of the cerebrovascular structure and flow and provide a reference for studying coordinated vascular changes across disease, aging, and intervention.