Automated Extraction of Morphological and Morphometric Characteristics of the Conjunctival Vasculature

Abstract

This report investigates a novel technique for delineating conjunctival vasculature from the scierai background. A Canny edge detector was used for accurate vessel edge detection, along with a “ravine finder” to define the axis of the vessel. A connectivity labeling algorithm was used to link adjacent pixels into strings in order to define vessel segments. This allows width measurement of numerous vessel segments, using appropriate triples of points, based on a central ravine point and two vessel boundary points. An initial investigation was conducted on 20 patients. Ages ranged from 20 to 65 years (mean=42.2, standard deviation=13.5 years). Results may indicate an increase in the number of larger vessels with age (ranging from 0.100 to 0.175 mm). Analysis shows a lognomial distribution of vessel widths. Potential applications of this technique are myriad. With further development of the connectivity algorithm individual vessels could be tracked allowing further morphometric and morphological characteristics to be quantified. This will only further the understanding of vascular “abnormality.”