Structural analysis of retinal vessels

We have investigated a method of iteratively tracking along blood vessels in a digitised image of the retina, recording information on the morphology of the individual vessels and their macrostructure. Such information may be of use in the detection of vessel changes in diabetes and hypertension, as well as be of use for further image analysis. The algorithm can successfully track along the vessel, detect and follow bifurcations and stop correctly at the end of the vessel. Currently, the algorithm is instigated manually, and further development is required to automate this process. The stopping criterion and bifurcation detection both depend on heuristically defined values, and adaptive knowledge based methods must be developed. The final goal of the algorithm is to delineate the entire vessel tree structure without user intervention.