PC-Seg: Progressive Cross-View Consistency for 3D OCT Segmentation from Sparse 2D Annotations
2026-07-20 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors developed PC-Seg, a new method to help computers recognize 3D structures in eye scan images using very few labeled examples. Instead of analyzing the images one slice at a time, their method looks at different views and makes the 2D and 3D models work together to improve accuracy. This approach uses only about 0.7% of the labeled data compared to usual methods but achieves similar accuracy to fully supervised learning. Tests on two eye disease datasets showed their method outperformed other semi-supervised approaches.
Volumetric segmentationOptical coherence tomography (OCT)Semi-supervised learning3D spatial contextB-scansPseudo-labelsCross-view consistencyCo-trainingRetinal layer segmentationDeep learning models
Authors
Tsubasa Konno, Takahiro Ninomiya, Yukun Zhou, Koichi Ito, Siegfried K. Wagner, Yiqun Lin, Pearse A. Keane, Toru Nakazawa, Takafumi Aoki
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.