6.1. 6.1 Blood Vessel Segmentation

In collaboration with Miguel Angel Palomera-Pérez

6.1.1. Problem Analysis Problem Statement

Retinal microvasculature measurement is coming into common use for diagnosis of several diseases that affect the retinal blood vessels. Some diseases, such as cardiovascular disorders or diabetes, can be diagnosed by analyzing a patient's retinal blood vessels. However, to form a precise diagnosis, retinal images must be segmented – that is, determining where in the image there is a blood vessel and where there is background retinal tissue. The more precise the blood vessel detection, the more useful is the application for diagnostic purposes. The large volume of retinal images makes the task of their processing and analysis a difficult and time-consuming one.

For example, consider the retinal image in Figure 6.1, which is taken from the STARE database [HKG00]. It is necessary to extract the features of this image, in terms of blood vessels, as accurately as possible. Such an accuracy is normally obtained by comparing with retinal images from public databases such as STARE [HKG00] and DRIVE [SAN+04].

In the example given here, the algorithm used for blood vessel segmentation is based on principles of multi-scale geometry for feature extraction, in combination with an iterative region-growing algorithm.

  • Feature extraction. Multi-scale geometry techniques allow information about the objects that compose a two-dimensional, two-color, retinal ...

Get Patterns for Parallel Software Design now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.