Reg-TTR, Test-Time Refinement for Fast, Robust and Accurate Image Registration

TL;DR

Reg-TTR tackles the trade-off between robustness and speed in medical image registration by combining a foundation-model prediction with test-time refinement. At inference, it first predicts an initial deformation field using a pre-trained network and then refines it through a small number of optimization steps, guided by a hybrid loss that promotes similarity and smoothness. The approach yields state-of-the-art–like accuracy on Abdomen CT and ACDC with inference times comparable to prior deep-learning methods, demonstrating strong generalization across tasks. By narrowing the gap between general registration foundation models and task-specific SOTA methods, Reg-TTR offers a practical and scalable solution for clinical workflows and can be extended to more datasets and modalities.

Abstract

Traditional image registration methods are robust but slow due to their iterative nature. While deep learning has accelerated inference, it often struggles with domain shifts. Emerging registration foundation models offer a balance of speed and robustness, yet typically cannot match the peak accuracy of specialized models trained on specific datasets. To mitigate this limitation, we propose Reg-TTR, a test-time refinement framework that synergizes the complementary strengths of both deep learning and conventional registration techniques. By refining the predictions of pre-trained models at inference, our method delivers significantly improved registration accuracy at a modest computational cost, requiring only 21% additional inference time (0.56s). We evaluate Reg-TTR on two distinct tasks and show that it achieves state-of-the-art (SOTA) performance while maintaining inference speeds close to previous deep learning methods. As foundation models continue to emerge, our framework offers an efficient strategy to narrow the performance gap between registration foundation models and SOTA methods trained on specialized datasets. The source code will be publicly available following the acceptance of this work.

Figures (4)

• Figure 1: The schema of our Reg-TTR. Reg-TTR framework operates in two stages: the first stage generates an initial deformation field using a pre-trained network, while the second stage further refines this field to achieve optimal registration performance. Reg-TTR effectively narrows the performance gap between general-purpose registration foundation models and data-specific models, thereby facilitating both rapid and accurate image registration.
• Figure 2: Qualitative comparison on the Abdomen CT dataset. uniGradICON + TTR achieves a better structure consistency.
• Figure 3: Quantitative comparison of 5 pre-trained models on the ACDC dataset before ('initial') and after ('optimized') applying the TTR optimization.
• Figure 4: Registration Dice scores of our Reg-TTR on ACDC and Abdomen CT datasets with increasing TTR iterations.