Spatio-temporal characterization of vessel segments applied to retinal angiographic images

In this paper, we present a new approach of analysis and recognition of retinal vessel segments for the quantification of their shapes change due to alterations. This approach is based on a spatial followed by a temporal characterization of retinal fluorescein angiographic images. The spatial characteristics (coordinates and classification number i.e. numbering of segments) are associated to bifurcation points (bp), which are matched in temporal image pairs for vessel segments correspondence forming. The matching process uses spatial and temporal characteristics between the bp and their surrounded vessel segments by computing a coefficient of similarity measurement. The recognition of vessel segments will help ophthalmologists in quantifying changes in vessel shape parameters and detect the temporal evolution of some retinal pathologies as the Sickle Cell retinopathy in our case. (C) 2002 Elsevier Science B.V. All rights reserved.