用稀疏专家路由分离视网膜多重病变，提升诊断可解释性。
Disentangling Co-Occurring Retinal Pathologies with Saliency-Guided Sparse Expert Routing
- 引入基于注意力的专家路由机制，按病灶特征动态分配计算路径。
- 在五类疾病上达到0.912宏AUC，且专家分配显著依赖于病理类型。
- 可视化证实专家分工与病灶位置一致，适合临床辅助诊断场景。
视网膜眼底图像常同时存在多种共现病变，而传统深度学习分类器对每张图像采用静态统一计算。本文提出一种新架构，通过稀疏条件计算实现解耦：前端使用引导上下文门控（GCG）空间注意力提取关键区域，后端基于特征令牌的稀疏路由混合专家（MoE）模块进行决策。该路由机制产生可解释的数据驱动分解，专家分配显著依赖于疾病类型（p < 0.001），健康状态及形态显著不同的病变（如ERM、AMD）被分配至专属专家。在五类疾病、患者独立的五折交叉验证基准上，模型取得0.912 ± 0.008宏AUC和0.653 ± 0.014宏F1。Grad-CAM++与后MoE t-SNE可视化表明，专家路由与局部病灶对齐，并在几何空间中映射出共现病例的组合关系，证明稀疏MoE是多疾病视网膜筛查中兼具性能与可解释性的有效方法。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。