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

2026-06-17

AI Summary

This work addresses the dimensional and domain discrepancies that arise when directly applying the natural-image pretrained model DINOv3 to 3D medical image segmentation. To bridge this gap, the authors propose a two-stage progressive adaptation framework: first, a multi-slice embedding module constructs pseudo-3D contextual representations, coupled with a proxy segmentation task to facilitate cross-domain transfer; second, a lightweight 3D adapter is introduced to enhance inter-slice global consistency, alongside a parallel high-frequency detail recovery stream to preserve boundary information. This approach represents the first successful and efficient adaptation of DINOv3 to 3D medical segmentation, achieving state-of-the-art performance across five public datasets while simultaneously maintaining strong semantic adaptability and spatial detail fidelity.

Abstract

Although DINOv3 has demonstrated remarkable semantic discrimination in natural imagery, its direct application to volumetric medical segmentation is hindered by inherent dimension and domain disparities. To resolve these issues, we propose DINO-Med3D, a two-stage progressive framework that repurpose the pre-trained DINOv3 encoder for 3D medical tasks. In the first stage, we mitigate the dimension gap by introducing a multi-slice embedding module that incorporates pseudo-3D context, while simultaneously employing a segmentation proxy task to adapt representations learned from natural scenes to the medical domain. Subsequently, we further enhance volumetric understanding by adding lightweight 3D adapters into the frozen backbone to enforce global inter-slice continuity. Finally, to compensate for the spatial information loss inherent in the embedding process, we design a parallel detail recovery stream to explicitly preserve high-frequency boundary cues. Extensive experiments on five public datasets demonstrate that our approach successfully adapts DINOv3 to the medical domain and significantly outperforms state-of-the-art baselines.

Problem

• volumetric segmentation
• domain gap
• dimension gap
• medical image analysis
• pre-trained vision models

Innovation

• progressive adaptation
• multi-slice embedding
• 3D adapters
• detail recovery stream
• domain transfer