Corneal ulcer is a common ophthalmic symptom. Segmentation algorithms are needed to identify and quantify corneal ulcers from ocular staining images. Developments of such algorithms have been obstructed by a lack of high quality datasets (the ocular staining images and the corresponding goldstandard ulcer segmentation labels), especially for supervised learning based segmentation algorithms. In such context, we prepare a dataset containing 712 ocular staining images and the associated segmentation labels of flaky corneal ulcers. In addition to segmentation labels for flaky corneal ulcers, we also provide each image with three-fold class labels: firstly, each image has a label in terms of its general ulcer pattern; secondly, each image has a label in terms of its specific ulcer pattern; thirdly, each image has a label indicating its ulcer severity degree. This dataset not only provides an excellent opportunity for investigating the accuracy and reliability of different segmentation and classification algorithms for corneal ulcers, but also advances the development of new supervised learning based algorithms especially those in the deep learning framework.

No abstract

No abstract

Although deep learning based diabetic retinopathy (DR) classification methods typically benefit from welldesigned architectures of convolutional neural networks, the training setting also has a non-negligible impact on the prediction performance. The training setting includes various interdependent components, such as objective function, data sampling strategy and data augmentation approach. To identify the key components in a standard deep learning framework (ResNet-50) for DR grading, we systematically analyze the impact of several major components. Extensive experiments are conducted on a publicly-available dataset EyePACS. We demonstrate that (1) the ResNet-50 framework for DR grading is sensitive to input resolution, objective function, and composition of data augmentation, (2) using mean square error as the loss function can effectively improve the performance with respect to a task-specific evaluation metric, namely the quadratically-weighted Kappa, (3) utilizing eye pairs boosts the performance of DR grading and (4) using data resampling to address the problem of imbalanced data distribution in EyePACS hurts the performance. Based on these observations and an optimal combination of the investigated components, our framework, without any specialized network design, achieves the state-of-the-art result (0.8631 for Kappa) on the EyePACS test set (a total of 42670 fundus images) with only image-level labels. Our codes and pre-trained model are available at https://github.com/YijinHuang/pytorch-classification.

No abstract