Imaging, image processing and pattern analysis of skin capillary ensembles

Jicun Zhong,Claes L. Asker,E. Göran Salerud

DOI: https://doi.org/10.1034/j.1600-0846.2000.006002045.x

Abstract:BACKGROUND/AIMS: The capillary bed is recognized as the site where metabolic and nutrient processes occur for living tissues at all levels. The evaluation of this vital process is a major concern in microcirculation. Unlike traditional approaches that concentrated on the extreme local properties of this process, a more global analysis toward capillary ensembles is employed here, since capillaries work as a cooperative entirety. As a first step toward ensemble analysis, the static and planar geometric parameters are investigated. Parameters such as the capillary adjacency and size information are very important in predicting and analysing certain malfunctions in the microvascular bed. METHODS/RESULTS: In order to achieve an objective and accurate analysis of these vital parameters, a computerized imaging system is proposed. Not only the number of capillaries and the capillary cross-sectional areas are important in describing the microvascular bed but the planar distribution pattern of the capillaries also carries valid information. This information, unique to the ensemble analysis, can be used to reveal, visualise and quantify the clustering of capillaries; and this information, according to the Krogh model, is fundamental in estimating the tissue oxygen supply. Two spatial models, the closest neighbor and triangulation methods, have been applied to the captured images of capillary ensembles. The closest neighbor technique generates a minimal distance map or displays a distribution, which depicts the local clustering of capillaries. The triangulation technique, on the other hand, generates a mutual distance map, which is a global description of the capillary positions. Triangulation methods have been evaluated but all except the Greedy triangulation method have been rejected due to lack of robustness and model weakness. Therefore, the capillaries are triangulated by the Greedy triangulation method, and the capillary distribution uniformity is defined as one minus the coefficient of variance of the edge lengths of the mutual distance map. CONCLUSIONS: A series of advanced image processing methods have been developed that efficiently extract the capillary position, size and distribution information from the images. These results facilitate the automatic counting of capillaries and the capillary size-related pathological analysis.