Üveite bağlı vitröz bulanıklığın otomatik fotografik derecelendirilmesi

In the presence of vitreous haze due to uveitis, fundus details become less discernible in fundus photography. Ophthalmologists use fundus photographs in order to evaluate the vitreous haze level of patients as an indication of the severity of the inflammation in the uveal tract. Manual inspection of fundus photographs is time consuming, in addition to being error prone. In this thesis, we propose an algorithmic methodology for automated photographic grading of vitreous haze. The proposed algorithm is founded on automatic quality assessment and fundus image segmentation techniques. The method has two stages, namely feature set extraction and classification. Several classifiers are trained using a data set of fundus images with varying degrees of vitreous haze. Promising results are presented based on a comprehensive set of extracted features and a support vector machine based classifier with a linear kernel function.