Automatic generation of glomerular capillary topological organization

Abstract

Glomerular structural changes are conventionally investigated by optical or electron microscopy on two-dimensional (2D) sections. To understand the relationship between functional and structural changes of glomerular capillary networks in more detail, three-dimensional (3D) investigation of the capillary tufts is required. Since confocal microscopy and scanning electron microscopy cannot completely show the 3D topological organization of the capillary tuft, we have developed an automatic method to obtain a 3D model of the glomerular capillary lumen structure and to derive its topological organization. Serial semithin sections of a glomerular tuft, from rat kidney tissue, were digitized at high resolution. Capillary lumens were digitally outlined and segmented images were automatically aligned. A 3D model of the capillary tuft was automatically generated using the Visualization Toolkit library and the Marching Cubes algorithm. We then developed an original algorithm for automatic 3D skeletonization of capillary lumen volume to identify capillary segments and bifurcations and to obtain the topological organization of the network and geometric parameters of capillary segments (length, radius, and spatial configuration). Capillary segment connectivity was graphically presented in a 2D layout with an automatic procedure, revealing the lobular organization of the network. This technique, successfully applied to serial sections of a glomerular capillary, can be used to study a population of glomerular capillaries to disclose the structural effects of pathological conditions. The methodology can be extended to other vascular structures, such as the microcirculation of neoplastic tissues.