Disentangling Co-Occurring Retinal Pathologies with Saliency-Guided Sparse Expert Routing
2026-08-10 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionMachine Learning
AI summaryⓘ
The authors developed a new deep learning model to analyze eye images that often show multiple eye diseases at once. Their model uses a special system that directs different parts of the image to expert neural networks based on the disease features, making it smarter and more focused. This system can also explain which expert handled which disease, improving understanding of how the model works. On tests, their approach performed well at identifying five eye conditions and showed that experts specialize in certain diseases. Visualizations confirmed that the model pays attention to important lesion areas and groups similar disease cases together.
Retinal fundus imagesDeep learning classifiersSparse conditional computationGuided Context Gating (GCG)Mixture-of-Experts (MoE)Spatial attentionGrad-CAM++t-SNE visualizationMulti-disease classificationCross-validation
Authors
Nagur Shareef Shaik, Jeongwoo Park, Yeong-Jin Kim, Jaeuk Jung, Hyunjung Oh, Dong Hye Ye
Abstract
Retinal fundus images frequently exhibit multiple co-occurring pathologies, yet standard deep learning classifiers apply static, identical computation to every image regardless of the underlying disease distribution. We propose a novel architecture that resolves this via sparse conditional computation, pairing a Guided Context Gating (GCG) spatial attention front-end with a sparsely-routed Mixture-of-Experts (MoE) block operating over feature tokens. Crucially, this routing yields an interpretable, data-driven decomposition. Expert allocation is significantly disease-dependent (p < 0.001), with the healthy Normal state and morphologically distinct pathologies (e.g., ERM, AMD) isolating to dedicated experts. On a five-class, patient-disjoint 5-fold cross-validation benchmark, our model achieves 0.912 +/- 0.008 macro AUC and 0.653 +/- 0.014 macro F1. Furthermore, Grad-CAM++ and post-MoE t-SNE visualizations confirm that expert routing aligns with localized lesions and geometrically maps co-occurring cases between their constituent clusters, positioning sparse MoE as an interpretable approach to multi-disease retinal screening.