Skip to content

Joint-embedding self-supervised representation learning for robust cone-beam CT segmentation.

Nov 2026 · Medical Image Analysis · Vol 116, pp. 104393 · 0 citations · 34 references
Medicine

Abstract

Cone-beam CT (CBCT) segmentation is challenged by reconstruction artifacts and limited reliable annotations. We investigate joint-embedding predictive architecture (JEPA)-style self-supervised learning for volumetric CBCT segmentation using CBCT-LGR, which extends our preliminary slice-wise conference study through fully three-dimensional local-global encoding, learnable semantic tokens, and a combined objective comprising local-global alignment, semantic alignment, and predictive consistency. We compare CBCT-LGR with supervised baselines (nnU-Net and STU-Net), generic self-supervised methods (BYOL-Med3D and Models Genesis), and foundation models (SAM, MedSAM, and Anatomix) using harmonized patient-level splits, preprocessing, and evaluation metrics while retaining each method's stated supervision setting. For representation-learning methods, frozen encoders are evaluated using supervised decoders trained with the same 10-subject labeled-data budget per dataset. Across three datasets, CBCT-LGR outperforms the generic self-supervised baselines on dataset-average overlap and boundary metrics. On CBCTLiTS, supervised and fine-tuned foundation-model approaches achieve higher absolute DSC under clean reconstruction conditions. Across five controlled reconstruction-quality levels, however, CBCT-LGR exhibits the smallest maximum relative DSC decline among all nine methods, with declines of 1.5% for liver and 14.5% for tumor, and achieves the highest numerical tumor DSC at the two most degraded levels. Results on Dataset 1 quantify agreement with imperfect reference annotations and do not establish true anatomical accuracy or a definitive method ranking. These findings define the capabilities and limitations of JEPA-style pretraining for volumetric CBCT segmentation.

View source

Similar papers

Sep 2026

Unsupervised pelvic CBCT-to-CT translation via cross-slice attention for 3D context modeling

CAMIT provides an effective trade-off between 2D efficiency and 3D contextual modeling, supporting treatment-day anatomical assessment and adaptive workflow guidance, and is evaluated on an unpaired pelvic CBCT/CT dataset.

Xu Chen, Yun-Kui Pang, J. Lian et al. · 0 citations
Open access Nov 2026

SHAPE-Net: structure-aware and confidence-guided learning for semi-supervised medical image segmentation

Remarkable progress has recently been made in semi-supervised medical image segmentation. Nevertheless, issues related to unreliable pseudo-labels and insufficient feature discriminability remain unresolved. To mitigate the aforementioned limitations, a multi-level collaborative structural perception learning network (...

Dong Shen, Tian-Tian Ji, Bin Shi et al. · 0 citations
#machine learning Preprint Oct 2026

L2GSSL: Label-Efficient Local-to-Global Self-Supervised Learning for Volumetric Healthcare Data

Healthcare institutions routinely accumulate high-dimensional computed tomography (CT) volumes, but only a fraction can be annotated by experts. We present L2GSSL, a local-to-global self-supervised learning framework that leverages unlabeled volumes by decomposing pretraining into intra-slice and inter-slice representa...

Po-Cheng Cheng, Hung-Hsuan Chen · 0 citations
Open access Nov 2026

Symmetry-Consistent Attention Supervision for Explainable Multi-label Temporal-Bone CT Assessment.

The purpose of this study is to develop an explainable deep learning framework for region-wise, multi-label assessment of inner- and middle-ear status on temporal-bone computed tomography (CT). The framework used a ResNet-18 encoder with class-specific attention pooling to predict eight labels representing normal and c...

Ying Li, Zheng Li, Shou-Qin Zhao et al. · 0 citations
Oct 2026

DINOSAM+: Structure-Guided Prompt Learning for Semi-Supervised 3D Medical Image Segmentation.

A frozen self-supervised representation model, DINOv2, is employed to extract coarse structural attention from volumetric slices, which is further refined by a lightweight structure-aware network into foreground, boundary, and background probability maps, thereby providing more discriminative structural cues for pseudo...

Si-Bo Qiao, Chaofei Yang, Zhi-Yuan Zhao et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.