PC-Seg: Progressive Cross-View Consistency for 3D OCT Segmentation from Sparse 2D Annotations

2026-07-20

AI Summary

This work addresses the high annotation cost in 3D optical coherence tomography (OCT) image segmentation by proposing a progressive cross-view consistency framework that achieves accurate volumetric segmentation using only sparse 2D annotations. The method employs a single 2D model to learn cross-view consistency between standard B-scans and their orthogonal slices, generating reliable 3D pseudo-labels. These pseudo-labels facilitate collaborative training between 2D and 3D models, enabling mutual refinement. By innovatively integrating cross-view consistency into semi-supervised OCT segmentation, the approach attains performance close to fully supervised methods while utilizing merely ~0.7% of the full annotation budget. Experiments on the MSHC and Duke DME datasets demonstrate significant improvements over existing semi-supervised and retinal layer segmentation techniques, effectively overcoming the limited exploitation of 3D contextual information inherent in conventional slice-by-slice processing.

Abstract

Volumetric segmentation of optical coherence tomography (OCT) images is essential for diagnosing ocular diseases but requires labor-intensive voxel-wise annotations. While semi-supervised learning (SSL) can reduce annotation costs, most existing methods process data slice by slice and fail to exploit the inherent 3D spatial context. We propose PC-Seg, a progressive cross-view consistency framework that learns high-accuracy 3D segmentation models from sparse 2D annotations. Unlike conventional multi-view approaches, PC-Seg uses a single 2D model to learn cross-view consistency from standard B-scans and orthogonal slices, thereby generating reliable volumetric pseudo-labels. These pseudo-labels are then distilled into a 3D model, followed by a co-training stage in which the 2D and 3D models mutually refine each other through ensemble pseudo-labeling. Experiments on the MSHC and Duke DME datasets demonstrate that PC-Seg achieves accuracy comparable to fully supervised learning while using labels for only about 0.7% of the training data, outperforming state-of-the-art semi-supervised and retinal layer segmentation methods. Our code is publicly available at https://github.com/gsisaoki/pc-seg-official.

Problem

• 3D OCT segmentation
• sparse 2D annotations
• semi-supervised learning
• cross-view consistency
• volumetric segmentation

Innovation

• progressive cross-view consistency
• 3D OCT segmentation
• sparse 2D annotations
• semi-supervised learning
• co-training