Automatic retinal vessel profiling using multi-step regression method

Caliber of the retinal blood vessel is widely used for risk assessment of cardiovascular diseases. Accurate and automatic caliber measurement requires a precise model to be made for the vessel profile. In this paper, we present a new approach for retinal vessel profiling in which the background noise, uneven illuminations and specular reflections have all been considered. In this method, regression analysis is performed with a series of second-order Gaussians to filter and up-sample the original vessel profile. This is then segmented to identify and represent the vessel edges by two Generalized Gaussian functions. The technique has been applied to retinal images and the results have been verified and compared with the state of the art automatic techniques.