Fully Convolutional Networks for Monocular Retinal Depth Estimation and Optic Disc-Cup Segmentation

This work addresses glaucoma diagnosis for medical professionals by providing automated tools for analyzing optic nerve head structure, though it appears incremental as it builds on existing convolutional network approaches.

The paper tackled the problem of glaucoma screening by developing a method for monocular retinal depth estimation and optic disc-cup segmentation from color fundus images, achieving results such as a 0.92 dice coefficient for segmentation and 0.15 mm mean absolute error for depth estimation.