Deep Convolutional Networks for Image Segmentation: Application to Optic Disc detection

There is various architecture of convolutional neural network (CNN) based on deep learning methods and techniques applied to modern computer vision tasks. However, models based on CNNs are used in image segmentation such as LeNet, FCN, AlexNet, VGGNet, GoogLeNet, and ResNet. In this paper we present the application of Deep CNN based model for Image Segmentation, especially to detect the Optic Disc from retina images. We use the Indian Diabetic Retinopathy Image Dataset (IDRiD), we also implement the model with keras in python. A brief description of model, training with data augmentation, validation and predictions is given, at the end we compare the result with an existing labeled data. Finally, an overview is given of future works of usage of the convolutional neural network for image segmentation problems in medical imaging context and the challenges involved therein.