Accurate segmentation of the optic disc (OD) depicted on color fundus images plays an important role in the earlydetection and quantitative diagnosis of retinal diseases, such as glaucoma and optic atrophy. In this study, we proposed acoarse-to-fine deep learning framework on the basis of a classical convolutional neural network (CNN), known as the Unetmodel, for extracting the optic disc from fundus images. This network was trained separately on fundus images andtheir vessel density maps, leading to two coarse segmentation results from the entire images. We combined the resultsusing an overlap strategy to identify a local image patch (disc candidate region), which was then fed into the U-netmodel for further segmentation. Our experiments demonstrated that the developed framework achieved an averageintersection over union (IoU) and a dice similarity coefficient (DSC) of 89.1% and 93.9%, respectively, based on a totalof 2,978 test images from our collected dataset and six public datasets, as compared to 87.4% and 92.5% obtained byonly using the sole U-net model. This suggests that the proposed method can provide better segmentation performancesand have the potential for population based disease screening.
展开▼