SD-RetinaNet: Topologically Constrained Semi-Supervised Retinal Lesion and Layer Segmentation in OCT

2025-09-25

AI Summary

Existing semi-supervised retinal OCT image segmentation methods often yield anatomically implausible results, fail to model layer–lesion interactions effectively, and lack topological guarantees. To address these limitations, we propose a joint layer-and-lesion topologically aware semi-supervised segmentation framework. Our method introduces a differentiable biomarker topology engine that enables bidirectional decoupled modeling of retinal layers and pathological lesions. It integrates spatial–style feature disentanglement, topology-aware regularization, and differentiable morphological operations to explicitly enforce anatomical consistency and topological correctness. Evaluated on multiple public and private OCT datasets, our approach achieves significant improvements over state-of-the-art methods. Notably, it maintains high segmentation accuracy and robust generalization to diverse pathologies even with only a small number of annotated samples.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Semi-Supervised Learning
• Search and Optimization: Learning to Search

Application Category

• Security and Privacy: Data transparency and provenance
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Optical coherence tomography (OCT) is widely used for diagnosing and monitoring retinal diseases, such as age-related macular degeneration (AMD). The segmentation of biomarkers such as layers and lesions is essential for patient diagnosis and follow-up. Recently, semi-supervised learning has shown promise in improving retinal segmentation performance. However, existing methods often produce anatomically implausible segmentations, fail to effectively model layer-lesion interactions, and lack guarantees on topological correctness. To address these limitations, we propose a novel semi-supervised model that introduces a fully differentiable biomarker topology engine to enforce anatomically correct segmentation of lesions and layers. This enables joint learning with bidirectional influence between layers and lesions, leveraging unlabeled and diverse partially labeled datasets. Our model learns a disentangled representation, separating spatial and style factors. This approach enables more realistic layer segmentations and improves lesion segmentation, while strictly enforcing lesion location in their anatomically plausible positions relative to the segmented layers. We evaluate the proposed model on public and internal datasets of OCT scans and show that it outperforms the current state-of-the-art in both lesion and layer segmentation, while demonstrating the ability to generalize layer segmentation to pathological cases using partially annotated training data. Our results demonstrate the potential of using anatomical constraints in semi-supervised learning for accurate, robust, and trustworthy retinal biomarker segmentation.

Problem

• Addresses anatomically implausible retinal layer and lesion segmentation in OCT
• Solves ineffective modeling of layer-lesion interactions in semi-supervised learning
• Ensures topological correctness for retinal biomarker segmentation accuracy

Innovation

• Differentiable topology engine enforces anatomically correct segmentation
• Joint learning with bidirectional influence between layers and lesions
• Disentangled representation separates spatial and style factors