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

Deep learning is currently the state-of-the-art for automated detection ofreferable diabetic retinopathy (DR) from color fundus photographs (CFP). Whilethe general interest is put on improving results through methodologicalinnovations, it is not clear how good these approaches perform compared tostandard deep classification models trained with the appropriate settings. Inthis paper we propose to model a strong baseline for this task based on asimple and standard ResNet-18 architecture. To this end, we built on top ofprior art by training the model with a standard preprocessing strategy butusing images from several public sources and an empirically calibrated dataaugmentation setting. To evaluate its performance, we covered multipleclinically relevant perspectives, including image and patient level DRscreening, discriminating responses by input quality and DR grade, assessingmodel uncertainties and analyzing its results in a qualitative manner. With noother methodological innovation than a carefully designed training, our ResNetmodel achieved an AUC = 0.955 (0.953 - 0.956) on a combined test set of 61007test images from different public datasets, which is in line or even betterthan what other more complex deep learning models reported in the literature.Similar AUC values were obtained in 480 images from two separate in-housedatabases specially prepared for this study, which emphasize its generalizationability. This confirms that standard networks can still be strong baselines forthis task if properly trained.

A ResNet is All You Need? Modeling A Strong Baseline for Detecting Referable Diabetic Retinopathy in Fundus Images