Real-time single-vessel analysis reveals vascular-type-dependent blood-brain barrier dysfunction in rodent models of...

Recent evidence highlights significant heterogeneity of the blood-brain barrier (BBB) across vascular types, which becomes more pronounced under disease conditions. However, functional changes in vascular-type-dependent BBB leakage remain poorly characterized.

We aimed to establish an analysis framework for quantifying single-vessel BBB total leakage and identifying vascular-type-dependent patterns of total leakage change in disease states.

We introduce a method that combines in vivo real-time two-photon imaging with uniform manifold approximation and projection (UMAP)-based dimensionality reduction to assess BBB total leakage at the single-vessel level. Two rodent models were used, pilocarpine-induced status epilepticus (SE) and lipopolysaccharide-induced neuroinflammation (NI), which exhibit differential pathophysiological characteristics of BBB impairment.

Real-time imaging clearly showed arterial BBB leakage in SE, whereas leakage in NI was likely venous. Conventional intensity-based metrics, including area under the curve (AUC), intensity fold change ( $\Delta F/F_0$ ), and averaged differential coefficient ( $\Delta F/\Delta t$ ), detected arteria-specific changes in SE but failed to capture vein-specific differences in NI. By contrast, UMAP-based analyses sensitively distinguished disease-specific total leakage patterns, allowing separation of SE from arterial data and NI from venous data.

This integrated approach enables quantitative evaluation of vascular-type-dependent BBB total leakage and provides a platform for future studies on vessel-specific BBB alterations in neurological disorders.