MPD$^2$-Router: Mask-aware Multi-expert Prior-regularized Dual-head Deferral Router in Glaucoma Screening and Diagnosis
2026-05-08 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors created MPD²-Router, a system that helps decide when a machine should handle glaucoma screening or defer to human experts, and which expert to choose based on their availability and expertise. Their method considers factors like image quality, uncertainty, and disease complexity, while managing expert workload and minimizing potential harm from mistakes. Tested on multiple datasets, their approach improved accuracy and lowered risks compared to AI alone, especially when data varied across sources. It also balanced the use of different experts to avoid overloading any single one.
learning-to-deferglaucoma screeninghuman-AI collaborationdeferral policyGumbel-sigmoid gatingasymmetric costcross-domain shiftexpert workload balancingmorphological featuresout-of-distribution detection
Authors
Wenxin Zhan
Abstract
Learning-to-defer (L2D) can make glaucoma screening safer by routing difficult/uncertain cases to humans, yet standard formulations overlook expert availability, heterogeneous readers behavior, workload imbalance, asymmetric diagnostic harm, case difficulty from morphology and deployment shift. We introduce MPD$^2$-Router, a mask-aware multi-expert deferral framework that recasts ophthalmic triage as constrained human--AI routing: whether to defer and to which available expert. It couples a dual-head deferral/allocation policy with mask-aware Gumbel--sigmoid gating that strictly enforces per-sample availability, and fuses uncertainty, morphology, image-quality, and OOD signals. Training uses an asymmetric cost-sensitive objective with an augmented-Lagrangian deferral budget, a group-specific distribution prior, and a rank-majorization JS regularizer that jointly prevent expert collapse without forcing uniform allocation. Across three cross-national glaucoma cohorts (REFUGE, CHAKSU, ORIGA) with a frozen REFUGE-trained backbone, MPD$^2$-Router substantially lowers clinical cost and improves MCC over AI-only at a moderate deferral rate. It is Pareto-optimal in F1--MCC--cost, robust under cross-domain shift, and yields balanced expert utilization.