UC San Diego

The vascular system maintains brain homeostasis optimized for the dynamic computation of neurons. The architecture of the vascular network is fundamental to its functionality as a vital transport system. Nonetheless, comprehensive studies into the structure of the entire brain vascular network and its role in homeostatic regulation have been limited, largely due to the multiscale complexity of the system. This thesis presents novel experimental and computational methodologies developed to construct and analyze mouse brain vascular connectome.

We developed techniques to completely label and image whole mouse brain vasculature at sub-micrometer resolution. An efficient computational pipeline was developed to convert immense raw data into a microvascular connectome, the spatial graph representation of the network documenting the position and radius of 6 million interconnected vessel segments in a trillion-voxel space, with 99.9% connectivity accuracy.

Utilizing this dataset, we analyzed the structure of the vascular network across brain regions. Topological analyses reveal a common network connection pattern across the brain, leading to a universal structural robustness rooted in percolation transition. Systematic quantification of network orientation preference reveals brain regions with striking microvascular anisotropy, which bears implications for interpreting functional magnetic resonant imaging (fMRI) data. By combining biophysical analysis with numerical simulations, we deduced a formula connecting resting-state metabolism rate to network density and further predicted a common value of maximum tissue oxygen tension across the brain. Extending beyond static structure, perturbation analyses quantified the impacts of single vessel dilation, constriction, and obstruction on local blood flow and tissue oxygenation.

Toward constructing vascular connectomes suitable for large-scale flow simulations, we further explored the use of nonlinear optical techniques to image the entire brain vasculature within the cranium at sub-micrometer resolution. High-resolution two-photon and second-harmonic imaging were combined with online processing to define ablation trajectories and parameters for different tissues. Spatiotemporally focused femtosecond pulses were applied for precise and efficient material removal. This entire process was automated through custom-built control software to ensure reliable multi-day operation. This system enabled a detailed examination of the complex vascular connection between the brain and the skull, vital for modeling cerebral blood flow.