🤖 AI Summary
This work addresses the limited physical interpretability and fine-grained deformation control in existing deep learning–based medical image registration methods. Inspired by the Demons algorithm, it introduces local discrepancy modeling as a physically grounded prior into the registration framework for the first time. A neural Demons layer generates anatomically consistent driving forces in latent feature space, while an attention mechanism coupled with variational registration enables dynamic force–displacement interaction. By decoupling discrepancy modeling from the deformation process, the approach enhances modularity and interpretability. Evaluated on multiple 3D brain MRI datasets, the method outperforms state-of-the-art learning-based and optimization-based approaches, with visualizations and statistical analyses confirming a high degree of alignment between the computed driving forces and actual deformations.
📝 Abstract
Medical image registration benefits significantly from deep learning, yet existing approaches often lack physical explainability and fine-grained deformation control. Motivated by Demons algorithms, we propose a novel DrivenMorph framework that bridges attention mechanisms with variational image registration by incorporating difference modeling as a physically inspired inductive bias. The resulting driving force, computed from local differences in the latent feature space, provides explicit semantic guidance throughout the registration process. It directly drives the registration process through a neural Demons layer that simulates force-displacement interactions to generate smooth and anatomically consistent deformation. Unlike previous methods, our approach not only integrates traditional registration principles with popular deep networks, providing an explainable and efficient solution for learning-based medical image registration, but also separates difference modeling from deformation, improving modularity and explainability. Extensive experiments on multiple 3D brain MRI datasets demonstrate superior performance over state of-the-art learning-based and optimization-based methods. Furthermore, visualizations and statistical analyses confirm that the learned driving force aligns closely with actual deformation patterns, supporting its explanatory value.