Identification of anatomic retinal structures for macular delineation in fluorescein angiograms

Often, in the clinical setting, the severity of eye diseases is graded according to the number of lesions and their spatial relation within the retina. Detection of the optic disc (OD) and fovea is an important task in retinal imaging due to its significance in both clinical and image understanding tasks. Moreover, spatial relations between these structures are used to define which of the two eyes is under examination and to delineate the region of the macula. To detect these structures, several approaches have been proposed, primarily based on brightness and shape information. Recently, approaches that combine brightness, shape and information regarding the vascular structures have been used with good results. Nevertheless, they have been used in the detection of the OD/fovea on normal fundus images and do not account for properties specific to fluorescein angiography (FA), such as intensity variations and the presence of lesions, which may complicate the application of the FA procedure. In this paper, a method which combines brightness information, namely, local mean intensity and local mean intensity variation, and geometric information from the major blood vessels is used in the simultaneous detection of the OD and fovea, and in delineating the macular region in FA. The benefits of the proposed combination-based approach, in terms of accuracy, are demonstrated by comparing the results with those of intensity-and shape-based techniques, when used on their own. The image set employed to evaluate the proposed technique consists of both healthy and unhealthy retinas. The techniques discussed in this contribution result to a robust, fast and accurate system for macular delineation, which is able to simultaneously detect the OD/fovea at different stages of the FA and also in the presence of lesions such as drusens and micro aneurysms.