Automatic Detection of Microaneurysms in OCT Images Using Bag of Features

Diabetic retinopathy (DR) caused by diabetes occurs as a result of changes in the retinal vessels and causes visual impairment. Microaneurysms (MAs) are early clinical signs DR, whose timely diagnosis can help detecting DR stages its development. It has been observed that MAs more common inner layers compared to outer eyes suffering from DR. Optical coherence tomography (OCT) is noninvasive imaging technique provides cross-sectional view retina, it used recent years diagnose many eye diseases. As result, this paper attempts identify areas with MA normal retina using OCT images. This work done dataset collected FA images 20 patients In regard, firstly fluorescein angiography (FA) were registered. Then, separated, features each these extracted Bag Features (BOF) approach Speeded-Up Robust Feature (SURF) descriptor. Finally, classification process was performed multilayer perceptron network. For criteria accuracy, sensitivity, specificity, precision, obtained results 96.33%, 97.33%, 95.4%, 95.28%, respectively. Utilizing detect automatically new idea, preliminary research field promising.