Diabetic retinopathy (DR) is a common eye ailment that can cause blindness if not detected early. In this study, a spike neural network model was trained and validated on a Kaggle dataset which consists of 4000 retinal images and uses microaneurysms for feature extraction with a double-ring filter to detect them. The SNN is the third-generation neural network, which resembles the biological neurons in the brain. The SNN which uses the multi STIP algorithm learns to recognize patterns in the images that are associated with diabetic retinopathy. The SNN is trained to classify new images as either having diabetic retinopathy or not and even detect the stages of diabetic retinopathy. The results on an available dataset show that the proposed SNN-based approach achieves 95% accuracy.