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

TL;DR

This paper introduces a domain-specific self-supervised pretraining method for 3D medical image registration, improving robustness and efficiency over existing models by learning specialized volumetric features.

Contribution

It demonstrates that DINO-style pretraining on medical images enhances registration performance, surpassing models trained on natural images and traditional registration methods.

Findings

Outperforms DINOv2 trained on natural images.

Requires less computational resources at inference.

Achieves superior accuracy in challenging interpatient abdominal registration.

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.