Multi-Scale Simulation of Cerebral Blood Flow and Oxygen Exchange for the Entire Mouse Brain

Abstract

Aging-related dementia affects a significant and growing population every year. To investigate novel intervention techniques, the relationship between the brain chemistry must be quantified in healthy and unhealthy brains. Neurovascular imaging has correlated age-related hypoxic tissue to plaque formation (amyloid-β) and changes in vascular structure. Imaging paradigms investigating this coupling (known as the neurovascular unit) have limited spatial domains (=imaging window and depth) accompanied by point-measurements of oxygen, blood flow, and vascular structure which are performed using different imaging modalities and even on different subjects. This fragmented information can be pieced together with computational methods (=simulations) but historically suffered from even smaller spatial domains and significant non-physiological modeling assumptions. The current work advances these computational methods, using topological properties of empirically-derived vascular structures to synthesize synthetic clones (increase sample size) and grow the entire cerebral circulation of the hemisphere in mouse. These synthetic networks are topologically equivalent to empirical reconstructions. Improvements in nonlinear solvability and analysis have allowed rigorous biphasic blood flow simulations on these structures (=red blood cells + plasma). These simulations have revealed a previously undescribed red blood cell gradient in the microcirculation where the red blood cell (RBC) volume fraction (hematocrit) increases deeper in the cortex than at the surface. This gradient may constitute an autoregulatory function of the brain that was previously unknown, where the higher concentration of RBCs brings more oxygen to the deeper brain tissues than plasma alone. Moreover, a novel method for coupling the vascular structure to the extravascular tissue has been created that is able to resolve the 3D structure of the microvessels throughout the simulation domain of the tissue without sacrificing mathematical solvability. This new paradigm has allowed simulation of oxygen tension with spatial domains significantly larger than previously simulated throughout the cortical microcirculation. These advancements can be used to simulate the neurovascular unit at the scale of the entire mouse brain, removing modeling assumptions at artificial boundary edges (at the edge of the imaging window). Preliminary results in predicting reduced oxygen tension in the aged brain are also offered.