Computer Aided Diagnosis and Out-of-Distribution Detection in Glaucoma Screening Using Color Fundus Photography

TL;DR

This paper presents a CNN-based approach for glaucoma screening using color fundus images, incorporating an energy-based OOD detection method to identify ungradable images, aiming for robustness in real-world scenarios.

Contribution

The study introduces an inference-time OOD detection technique combined with CNN classification for glaucoma screening, enhancing robustness to ungradable images.

Findings

Effective CNN classification of glaucoma from fundus images

Successful integration of energy-based OOD detection

Improved robustness to ungradable images in screening

Abstract

Artificial Intelligence for RObust Glaucoma Screening (AIROGS) Challenge is held for developing solutions for glaucoma screening from color fundus photography that are robust to real-world scenarios. This report describes our method submitted to the AIROGS challenge. Our method employs convolutional neural networks to classify input images to "referable glaucoma" or "no referable glaucoma". In addition, we introduce an inference-time out-of-distribution (OOD) detection method to identify ungradable images. Our OOD detection is based on an energy-based method combined with activation rectification.