| dc.description.abstract | One of the microvascular complications caused by diabetes is diabetic retinopathy (DR). DR is damage to blood vessels in the area of the retina of the eye which causes swelling to leak fluid in the eye. Eye damage due to DR is the fourth most common disease in the world as a cause of vision loss or blindness. Early examination is important for people with DR to minimize blindness. One of the modern techniques for examining damage to the inside of the retina caused by DR (Diabetic Retinopathy) is using Optical Coherence Tomography (OCT). Currently, manual OCT is prone to speckle noise because it has low image contrast between retinal layers, making it difficult to distinguish retinal anatomy and influencing the diagnosis of DR disease. Therefore, in the prediction of DR disease through OCT images, pattern processing is required fast, precise and accurate machine learning. One method that can be used in OCT image pattern processing is Probabilistic Neural Network (PNN). The system design stage begins with doing preprocessing that consists of resizing, retinal segmentation, aignment and daubechies wavalet used as feature extraction. The PNN algorithm is used to identify OCT images by applying a trained model. The use of the PNN method is successful, where the system is able to identify DR and Normal diseases with good accuracy of 99%. | en_US |
