Two for One: Privileged Bilateral Learning for Monocular Inference in Individuals with Disabilities

Machine learning systems are often developed under data-rich conditions but deployed when only incomplete information is available. A natural example arises in colour fundus photography (CFP), where both eyes are commonly imaged because they can provide complementary information for ocular disease assessment. However, bilateral acquisition may not always be feasible, requiring models to operate from a single-eye image at inference. We propose a privileged learning framework in which a teacher model has access to bilateral CFP images during training and learns representations informed by information from both eyes. A student model is then trained to transfer this knowledge from the bilateral teacher while receiving only a single-eye image. At inference, the student operates exclusively on this single-eye input and does not require the contralateral image. Experimental results show consistent improvements in area under the receiver operating characteristic curve (AUROC) over conventional monocular training, with the largest gains observed for left-eye images for age-related macular degeneration (0.931 vs. 0.966) and diabetic retinopathy (0.789 vs. 0.836), while requiring only a single-eye image at inference. By enabling models to benefit from bilateral information during training while requiring only a single-eye image at inference, our approach may support more reliable deployment in resource-constrained clinical and screening environments.