Automated micro aneurysm classification using deep convolutional spike neural networks

One of the common diseases in people with micro aneurysms is diabetic retinopathy (DR). Due to a lack of early diagnosis, diabetic retinopathy poses a risk to vision because it develops without any warning symptoms. Therefore, the deep learning methods on color fundus images demonstrate the recognition task of diabetic retinopathy levels. In this manuscript, an automated Micro aneurysm detection and classification utilizing deep convolution spike neural network (DCSNN-AMA) is proposed for diabetic retinopathy. Here, the images are filtered using Savitzky-Golay (SG) pre-processing method. After preprocessing, the images are segmented under Tsalli's Entropy based multilevel 3D Otsu (TE-3D-Otsu) thresholding technique. The images are extracted under Gray level co-occurrence matrix (GLCM) window adaptive algorithm operation. After that, the Deep Convolutional Spiking Neural Network (DCSNN) method is employed for classification. The proposed method has attained 32.5, 19 and 24.4% higher accuracy, 23, 31.6 and 24.4% higher F-measure, and 19.35, 8 and 16.12% lower computation time analyzed to the existing models.