English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

Although deep learning based diabetic retinopathy (DR) classification methodstypically benefit from well-designed architectures of convolutional neuralnetworks, the training setting also has a non-negligible impact on theprediction performance. The training setting includes various interdependentcomponents, such as objective function, data sampling strategy and dataaugmentation approach. To identify the key components in a standard deeplearning framework (ResNet-50) for DR grading, we systematically analyze theimpact of several major components. Extensive experiments are conducted on apublicly-available dataset EyePACS. We demonstrate that (1) the DR gradingframework is sensitive to input resolution, objective function, and compositionof data augmentation, (2) using mean square error as the loss function caneffectively improve the performance with respect to a task-specific evaluationmetric, namely the quadratically-weighted Kappa, (3) utilizing eye pairs booststhe performance of DR grading and (4) using data resampling to address theproblem of imbalanced data distribution in EyePACS hurts the performance. Basedon these observations and an optimal combination of the investigatedcomponents, our framework, without any specialized network design, achieves thestate-of-the-art result (0.8631 for Kappa) on the EyePACS test set (a total of42670 fundus images) with only image-level labels. We also examine the proposedtraining practices on other fundus datasets and other network architectures toevaluate their generalizability. Our codes and pre-trained model are availableat https://github.com/YijinHuang/pytorch-classification.

Identifying the key components in ResNet-50 for diabetic retinopathy grading from fundus images: a systematic investigation