DualStreamFoveaNet: A Dual Stream Fusion Architecture with Anatomical Awareness for Robust Fovea Localization
This work addresses the challenge of accurately localizing the fovea in retinal images, particularly for diseased cases, which is important for medical professionals in ophthalmology to aid in disease analysis and prevention of vision loss.
The paper tackled the problem of robust fovea localization in retinal images, which is crucial for analyzing diseases to prevent vision loss, by proposing DualStreamFoveaNet, a transformer-based architecture that achieves state-of-the-art performance on multiple datasets and demonstrates improved robustness and generalization.
Accurate fovea localization is essential for analyzing retinal diseases to prevent irreversible vision loss. While current deep learning-based methods outperform traditional ones, they still face challenges such as the lack of local anatomical landmarks around the fovea, the inability to robustly handle diseased retinal images, and the variations in image conditions. In this paper, we propose a novel transformer-based architecture called DualStreamFoveaNet (DSFN) for multi-cue fusion. This architecture explicitly incorporates long-range connections and global features using retina and vessel distributions for robust fovea localization. We introduce a spatial attention mechanism in the dual-stream encoder to extract and fuse self-learned anatomical information, focusing more on features distributed along blood vessels and significantly reducing computational costs by decreasing token numbers. Our extensive experiments show that the proposed architecture achieves state-of-the-art performance on two public datasets and one large-scale private dataset. Furthermore, we demonstrate that the DSFN is more robust on both normal and diseased retina images and has better generalization capacity in cross-dataset experiments.