A Multi-resolution Entropic-Based Image Processing Technique for Diagnostic Analysis of Microcirculation Videos

Segmentation of microcirculation is a major step in analyzing blood circulatory system. In this study, a method for segmentation of capillaries and small blood vessels in video recordings captured from lingual surface is presented. The main objective is to monitor and quantitatively assess the variations that occur in microcirculation over a period of treatment for diseases or during a resuscitation process. The resulting system will assist physicians and researchers to make diagnostically and therapeutically important decisions such as determination of effectiveness of drug usage and resuscitation process. The algorithm is based on advanced digital image processing methods to detect capillaries and to determine the Functional Capillary Density (FCD) in order to identify the distribution of blood flow for a variety of normal and abnormal cases. The technique consists of four main steps; preprocessing, video stabilization, segmentation and post-processing. This study specifically focuses on minimizing user interaction and improving the accuracy of results. Visual inspection of the results by medical experts indicates that the technique can identify 95% of active capillaries and small blood vessels.