Unsupervised Domain Adaptation with Target-Only Margin Disparity Discrepancy

Gauthier Miralles; Loïc Le Folgoc; Vincent Jugnon; Pietro Gori

cs.CVMar 10, 2026

Unsupervised Domain Adaptation with Target-Only Margin Disparity Discrepancy

Authors: Gauthier Miralles, Loïc Le Folgoc, Vincent Jugnon, Pietro Gori

Organizations: LTCI, Télécom Paris, Institut Polytechnique de Paris, Palaiseau, France · GE Healthcare, Buc, France

Abstract

In interventional radiology, Cone-Beam Computed Tomography (CBCT) is a helpful imaging modality that provides guidance to practicians during minimally invasive procedures. CBCT differs from traditional Computed Tomography (CT) due to its limited reconstructed field of view, specific artefacts, and the intra-arterial administration of contrast medium. While CT benefits from abundant publicly available annotated datasets, interventional CBCT data remain scarce and largely unannotated, with existing datasets focused primarily on radiotherapy applications. To address this limitation, we leverage a proprietary collection of unannotated interventional CBCT scans in conjunction with annotated CT data, employing domain adaptation techniques to bridge the modality gap and enhance liver segmentation performance on CBCT. We propose a novel unsupervised domain adaptation (UDA) framework based on the formalism of Margin Disparity Discrepancy (MDD), which improves target domain performance through a reformulation of the original MDD optimization framework. Experimental results on CT and CBCT datasets for liver segmentation demonstrate that our method achieves state-of-the-art performance in UDA, as well as in the few-shot setting.

Figures & tables

Figure 2: Our feature-alignment UDA for segmentation. A U-Net is decomposed into a feature extractor $\psi$ , and a segmentation head $f$ . An adversary $f^{\prime}$ is built as a duplication of $f$ for adversarial training during UDA. After UDA, the features $z^{S}=\psi(x^{S})$ and $z^{T}=\psi(x^{T})$ are supposed to be domain-invariant, and $f^{\prime}$ is removed at inference. The loss $\mathcal{L}_{task}$ is the supervision segmentation loss, and $\mathcal{L}_{CE}$ is used for UDA.

Type	Method	F1 (%) $\uparrow$
Fully supervised	Source Only	54.1
	Target Only (1 vol)	54.3 $\pm$ 3.7
	Target Only (5 vol)	68.1 $\pm$ 3.9
	Target Only (20 vol)	77.4 $\pm$ 0.6
	Target Only (50 vol)	81.4 $\pm$ 0.5
	Target Only (100%)	85.5

Table 1: 2D CT to CBCT segmentation F1 scores for various methods. Best results are highlighted in bold . Standard deviations are reported for experiments repeated 5 times due to randomness in selecting a point for foundation model prompts or randomly choosing slices in few-shot settings.

Type	Method	F1 (%) $\uparrow$
Fully supervised	Source Only	80.1
	Target Only (1 vol)	61.0 $\pm$ 1.7
	Target Only (5 vol)	84.7 $\pm$ 0.9
	Target Only (20 vol)	89.6 $\pm$ 0.6
	Target Only (100%)	93.7
Foundation model	SAM-MED 3D [ 8 ] (1 pt)	53.6 $\pm$ 0.3

Table 2: 3D CT to CBCT segmentation F1 scores for various methods. Best results are highlighted in bold . Standard deviations are reported for experiments repeated 5 times due to randomness in selecting a point for foundation model prompts or randomly choosing scans in few-shot settings.

Figure 3: Segmentation results from various 3D methods. Ground truth mask is shown in red, and predicted segmentations in blue. High-intensity regions within the liver that cause the networks to fail are marked with yellow arrows. The same axial slice is displayed for each method with the corresponding prediction overlaid. The SAM-MED 3D prediction was generated using 5 randomly sampled points. The F1 score above each method, reported as a percentage, represents the 3D F1 score computed across the entire volume.

Figure 4: Stability Analysis of Hyperparameters in 2D CT to CBCT Unsupervised Domain Adaptation (UDA). This figure reports the F1 scores obtained on the target domain test set for different combinations of the hyperparameters $\alpha$ and $\gamma$ , with F1 values presented as percentages. The consistent performance across varying values of $\alpha$ and $\gamma$ confirms the robustness and stability of our method with respect to these hyperparameters.

Explore similar work

Unsupervised Adaptation of 3D CT Foundation Models for 3D CBCT Segmentation
Aug 27, 2026Gauthier Miralles, Loic Le Folgoc, Vincent Jugnon +1Cross-Modality Medical Image Segmentation 3D Medical Image Segmentation
LTCI, Télécom Paris, Institut Polytechnique de Paris, Palaiseau, France · GE HealthCare, Buc, France
BCMDA: Bidirectional Correlation Maps Domain Adaptation for Mixed Domain Semi-Supervised Medical Image Segmentation
Mar 25, 2026Bentao Song, Jun Huang, Qingfeng WangSemi-Supervised Medical Image Segmentation Unsupervised Domain Adaptation
School of Computer Science and Technology, Southwest University of Science and Technology, Mianyang, Sichuan, 621010, China
Unified CT and MRI Pancreas Segmentation for Label-Efficient Cross-Modality Subregion Transfer
Sep 11, 2026Ziliang Hong, Hongyi Pan, Halil Ertugrul Aktas +7Domain-Adversarial Learning Cross-Modality Medical Image Segmentation
Department of Radiology, Northwestern University, Chicago, IL, USA · Division of Gastroenterology and Hepatology, Mayo Clinic Florida, Jacksonville, FL, USA · Department of Gastroenterology and Hepatology, Northwestern University, Chicago, IL, USA

Aug 27, 2026J/K move · Enter open · S save

Unsupervised Adaptation of 3D CT Foundation Models for 3D CBCT Segmentation

Gauthier Miralles, Loic Le Folgoc, Vincent Jugnon, Pietro Gori

LTCI, Télécom Paris, Institut Polytechnique de Paris, Palaiseau, France · GE HealthCare, Buc, France

Cross-Modality Medical Image Segmentation 3D Medical Image Segmentation Unsupervised Domain Adaptation Vision Foundation Model Adaptation Cone-Beam CT

Open paper AskPDF arXiv

Accurate 3D segmentation of cone-beam CT (CBCT) is critical for interventional and radiation therapy applications, yet it remains limited by two compounding challenges: the scarcity of annotated CBCT data and the large domain shift from diagnostic CT. Interventional CBCT exhibits fundamental modality differences from conventional CT, driven by acquisition and physics effects as well as contrast-specific vascular content, thereby limiting effective cross-modality model transfer. We propose a novel unsupervised domain adaptation (UDA) framework based on redundancy-reducing feature alignment, enabling 3D CBCT segmentation with no target-domain annotations or inference-time adaptation. Our framework is architecture-agnostic, seamlessly adapting both CNN-based and ViT-based foundation models. We evaluate our method on two challenging CT-CBCT liver segmentation benchmarks: one for interventional vascular procedures and one for radiation therapy, demonstrating that even large-scale pretrained segmentation networks require explicit feature-space bridging to generalize across acquisition modalities, and that our approach consistently outperforms existing pretrained foundation model and UDA strategies. To support reproducibility and benchmarking, we release the liver segmentations for a public CBCT dataset, along with the code, trained models, and weights.

Abstract

Explore similar work

Unsupervised Adaptation of 3D CT Foundation Models for 3D CBCT Segmentation

BCMDA: Bidirectional Correlation Maps Domain Adaptation for Mixed Domain Semi-Supervised Medical Image Segmentation

Unified CT and MRI Pancreas Segmentation for Label-Efficient Cross-Modality Subregion Transfer