Arteriolar-to-venular diameter ratio estimation: A pixel-parallel approach

Abstract

The study of blood vessel features plays an important role in order to characterise markers used in early disease diagnosis. The arteriolar-to-venular (AVR) diameter ratio is an earlier marker related with cardiovascular risk, hypertension and diabetes. The extraction of the retinal vessel tree is not only the main task related with those medical applications intended to compute the AVR ratio, but it also implies a high computation effort. From the image processing point of view, many strategies and algorithms have been developed to deal with the extraction of this retinal vessel tree but specially regarding on the accuracy, but the execution time is still an open problem. In this paper, a methodology to extract the retinal vessel tree, tested in a fine-grain pixel-parallel processor array, is integrated into an application for the estimation of the AVR ratio in angiographies.

FPGA-accelerated retinal vessel-tree extraction

“…Nevertheless, the advent of sub-micrometer CMOS technologies leads to compact, fast and low-power hardware solutions. Person authentication systems based on the skeleton or some features extracted from the geometrical structure of the retinal vessel tree [1], fast and reliable arteriolar to venular diameter ratio estimation for diagnostic purpose [2], or even very fast retinal image registration as initial step towards real time video processing [3] are some examples of application which can benefit from hardware implementations.…”