Abstract:
Retino CAD is a Mobile Pan Optic Ophthalmoscope based automated decision support system designed for the detection of early symptoms of Proliferative Diabetic Retinopathy (PDR).
Proliferative diabetic retinopathy (PDR), an advanced stage of Diabetic Retinopathy, is a condition that carries a high risk of severe visual impairment. The hallmark of PDR is neovascularization, the growth of abnormal new vessels. Retinal image is acquired through Ophthalmoscope which leads to two parallel processing sequences; one of them drives to a procedure that identifies both new blood vessels and normal vessels and the other one reduces false responses to non-vessel edges. Both generated binary vessel maps, hold vital information which must be processed separately. Local morphology features are measured from each binary vessel map to produce two separate feature sets. Dual classification is performed on both feature sets independently using Ensemble Bagged/Boosted Decision Trees. The system then combines these individual classification outcomes through majority voting and determines the presence of PDR.
Retino CAD is especially suited for all diabetic patients to prevent them from vision loss by detecting the symptoms of PDR beforehand reducing the manual grading workload with the capability of detecting the presence of new vessels whilst reducing false responses.
