Glomerular Haemodynamics

Sammanfattning

Understanding the three-dimensional blood flow and fluid filtration within the renal glomerulus is essential for studying kidney function. However, directly measuring these local microvascular physical parameters remains difficult using traditional in vivo experimental techniques. This thesis presents a computational pipeline designed to transform high-resolution confocal microscopy data from a mouse kidney into a functional 3D finite element model to simulate glomerular haemodynamics.Using a combination of tissue clearing protocols and fluorescent labels, high-contrast image stacks of a mouse glomerulus were successfully captured. A fully connected 3D capillary lumen volume was isolated using machine learning segmentation combined with manual corrections. To satisfy computational mesh and workstation hardware constraints, the segmented volume was simplified into a network of smooth, straight cylinders.Computational fluid dynamics and mass transport equations were coupled and solved within COMSOL Multiphysics. The simulation successfully replicated the continuous hydrostatic pressure decline from the afferent to the efferent arteriole, mapped distinct zones of near-identical pressure suggesting the existence of capillary lobules, and captured the localised rise in oncotic pressure driven by plasma protein concentration. However, the baseline flow magnitudes were higher than standard literature values. Furthermore, data analysis revealed a mismatch between the model and biology, suggesting that the constriction and dilation abilities of the afferent and efferent arterioles are essential for correct glomerular function.Ultimately, this thesis establishes a specific, verifiable 3D framework that enables a scalable estimation of localised glomerular pressure fields and filtration gradients, offering a foundational tool for renal studies.