Effective Feature Learning for 3D Medical Registration via Domain-Specialized DINO Pretraining

Eytan Kats; Mattias P. Heinrich

cs.CVMar 14, 2026

Effective Feature Learning for 3D Medical Registration via Domain-Specialized DINO Pretraining

Authors: Eytan Kats, Mattias P. Heinrich

Organizations: Institute of Medical Informatics, University of Lübeck, Germany

Abstract

Medical image registration is a critical component of clinical imaging workflows, enabling accurate longitudinal assessment, multi-modal data fusion, and image-guided interventions. Intensity-based approaches often struggle with interscanner variability and complex anatomical deformations, whereas feature-based methods offer improved robustness by leveraging semantically informed representations. In this work, we investigate DINO-style self-supervised pretraining directly on 3D medical imaging data, aiming to learn dense volumetric features well suited for deformable registration. We assess the resulting representations on challenging interpatient abdominal registration task across both MRI and CT modalities. Our domain-specialized pretraining outperforms the DINOv2 model trained on a large-scale collection of natural images, while requiring substantially lower computational resources at inference time. Moreover, it surpasses established registration models under out-of-domain evaluation, demonstrating the value of task-agnostic yet medical imaging-focused pretraining for robust and efficient 3D image registration.

Figures & tables

Figure 1: Zero-shot registration framework. Features from the moving and fixed images are first extracted using the pretrained transformer and then projected into a low-dimensional space using PCA. ConvexAdam leverages these semantically rich representations to estimate an accurate displacement field.

Figure 2: Comparison of registration performance across different feature extractors. The results highlight the advantage of domain-specific DINO-style pretraining over DINOv2 and MIND features.

patch size	$DSC$	$SD\log J$	$\|J\|_{<0}$
$12\times 12\times 12$	${49.09}_{\pm 15.72}$	$\textbf{0.264}_{\pm 0.049}$	$\textbf{0.01}_{\pm 0.05}$
$8\times 8\times 8$	${51.02}_{\pm 17.21}$	$0.269_{\pm 0.059}$	$\textbf{0.01}_{\pm 0.04}$
$4\times 4\times 4$	$\textbf{57.74}_{\pm 20.14}$	$0.388_{\pm 0.099}$	$0.11_{\pm 0.2}$

Table 1: Impact of token resolution on downstream registration task. Increasing token resolution yields consistent performance gains, highlighting its role in capturing fine anatomical details.

	AbdomenCTCT			NAKO
method	$DSC$	$SD\log J$	$\|J\|_{<0}$	$DSC$	$SD\log J$	$\|J\|_{<0}$
Initial	$25.31_{\pm 6.89}$	-	-	$29.57_{\pm 14.66}$	-	-
MIND [ 6 ]	$37.08_{\pm 9.1}$	$\textbf{0.451}_{\pm 0.092}$	$\textbf{0.29}_{\pm 0.54}$	$50.55_{\pm 20.6}$	$0.398_{\pm 0.112}$	$0.11_{\pm 0.3}$
DINOv2 [ 8 ]	$34.73_{\pm 7.99}$	$0.485_{\pm 0.106}$	$0.67_{\pm 1.25}$	$54.95_{\pm 16.07}$	$\textbf{0.326}_{\pm 0.085}$	$\textbf{0.01}_{\pm 0.03}$
TotalRegistrator [ 9 ]	36.2	-	-	-	-	-
UniGradIcon [ 13 ]	34.1	-	0.02	-	-	-

Table 2: Registration results on the AbdomenCTCT and NAKO datasets. The findings demonstrate the benefit of domain-specific DINO-style pretraining for dense deformable registration. Reported values for TotalRegistrator and UniGradIcon are taken from their respective publications and correspond to out-of-domain evaluations, where the models were not trained on AbdomenCTCT.

Explore similar work

CoRe: Joint Optimization with Contrastive Learning for Medical Image Registration
Mar 24, 2026Eytan Kats, Christoph Grossbroehmer, Ziad Al-Haj Hemidi +3Medical Image Registration Deformable Image Registration
Insitute of Medical Informatics, University of Luebeck, Luebeck, Germany
DINO-Med3D: Bridging Dimension and Domain Gaps in Volumetric Segmentation via Progressive Adaptation
Jun 17, 2026Haoyu Hu, Xiyao Ma, Shiqi Liu +43D ViTs Domain Adaptation
University of Chinese Academy of Sciences, Beijing, China · Institute of Automation, Chinese Academy of Sciences, Beijing, China
Ordered Diffusion for 3D Human Registration
Aug 6, 2026Mattia Masiero, Ilya A. Petrov, Daniel Cremers +2Point Cloud Registration 3D Diffusion Models
University of Tübingen, Germany · Tübingen AI Center, Germany · Technical University of Munich, Germany +2

Mar 24, 2026J/K move · Enter open · S save

CoRe: Joint Optimization with Contrastive Learning for Medical Image Registration

Eytan Kats, Christoph Grossbroehmer, Ziad Al-Haj Hemidi, Fenja Falta+2

Insitute of Medical Informatics, University of Luebeck, Luebeck, Germany

Medical Image Registration Deformable Image Registration Medical Image Representation Learning Contrastive Learning

Open paper AskPDF arXiv

Medical image registration is a fundamental task in medical image analysis, enabling the alignment of images from different modalities or time points. However, intensity inconsistencies and nonlinear tissue deformations pose significant challenges to the robustness of registration methods. Recent approaches leveraging self-supervised representation learning show promise by pre-training feature extractors to generate robust anatomical embeddings, that farther used for the registration. In this work, we propose a novel framework that integrates equivariant contrastive learning directly into the registration model. Our approach leverages the power of contrastive learning to learn robust feature representations that are invariant to tissue deformations. By jointly optimizing the contrastive and registration objectives, we ensure that the learned representations are not only informative but also suitable for the registration task. We evaluate our method on abdominal and thoracic image registration tasks, including both intra-patient and inter-patient scenarios. Experimental results demonstrate that the integration of contrastive learning directly into the registration framework significantly improves performance, surpassing strong baseline methods.

Abstract

Explore similar work

CoRe: Joint Optimization with Contrastive Learning for Medical Image Registration

DINO-Med3D: Bridging Dimension and Domain Gaps in Volumetric Segmentation via Progressive Adaptation

Ordered Diffusion for 3D Human Registration