Automatic generation of glomerular capillary topological organization

Glomerular structural changes are conventionally investigated by optical or electron microscopy on two-dimensional (2D) sections. To understand the re lationship between functional and structural changes of glomerular capillar y networks in more detail, three-dimensional (3D) investigation of the capi llary tufts is required. Since confocal microscopy and scanning electron mi croscopy cannot completely show the 3D topological organization of the capi llary tuft, we have developed an automatic method to obtain a 3D model of t he glomerular capillary lumen structure and to derive its topological organ ization. Serial semithin sections of a glomerular tuft, from rat kidney tis sue, were digitized at high resolution. Capillary lumens were digitally out lined and segmented images were automatically aligned. A 3D model of the ca pillary tuft was automatically generated using the Visualization Toolkit li brary and the Marching Cubes algorithm. We then developed an original algor ithm for automatic 3D skeletonization of capillary lumen volume to identify capillary segments and bifurcations and to obtain the topological organiza tion of the network and geometric parameters of capillary segments (length, radius, and spatial configuration). Capillary segment connectivity was gra phically presented in a 2D layout with an automatic procedure, revealing th e lobular organization of the network. This technique, successfully applied to serial sections of a glomerular capillary, can be used to study a popul ation of glomerular capillaries to disclose the structural effects of patho logical conditions. The methodology can be extended to other vascular struc tures, such as the microcirculation of neoplastic tissues. (C) 2001 Academi c Press.