Blood Capillary Length Estimation from Three-Dimensional Microscopic Data by Image Analysis and Stereology

Abstract

AbstractStudies of the capillary bed characterized by its length or length density are relevant in many biomedical studies. A reliable assessment of capillary length from two-dimensional (2D), thin histological sections is a rather difficult task as it requires physical cutting of such sections in randomized directions. This is often technically demanding, inefficient, or outright impossible. However, if 3D image data of the microscopic structure under investigation are available, methods of length estimation that do not require randomized physical cutting of sections may be applied. Two different rat brain regions were optically sliced by confocal microscopy and resulting 3D images processed by three types of capillary length estimation methods: (1) stereological methods based on a computer generation of isotropic uniform random virtual test probes in 3D, either in the form of spatial grids of virtual “slicer” planes or spherical probes; (2) automatic method employing a digital version of the Crofton relations using the Euler characteristic of planar sections of the binary image; and (3) interactive “tracer” method for length measurement based on a manual delineation in 3D of the axes of capillary segments. The presented methods were compared in terms of their practical applicability, efficiency, and precision.