Rapid, Automated, and Robust Tracing of Vasculature, and Quantification of Visual Features from Angiographic Images

Abstract

The rapid tracing of vasculature, and quantification of visual landmarks such as those corresponding to the branching and crossover points from images is extremely useful in a variety of fields including ophthalmology, radiology, and neurobiology. A novel algorithm that enables rapid, automatic, robust, adaptive, and accurate tracing of vasculature has been developed. Given an image (Fig 1), it automatically generates a labeled graph representation of the vasculature (Fig 2), and a table listing the locations of visual landmarks such as crossover and branching points (Fig 3). For each such point, it also lists the intersection angles, and the local thickness of the intersecting vasculature segments. This data is often useful by itself. They can also be used for image registration and montage building.This algorithm greatly improves upon previous work in several ways. First, it is fully automatic, and free of parameters that need to be adjusted carefully (just needs optional sensitivity setting).