DARE: A Deformable Adaptive Regularization Estimator for Learning-Based Medical Image Registration

2025-10-22

AI Summary

In medical image registration, existing deep learning methods often suffer from rigid regularization designs, leading to non-robust deformations, anatomically implausible alignments, and folding artifacts. To address these issues, we propose DARE—a novel framework featuring learnable dynamic elastic regularization that adaptively balances strain and shear energy weights based on the gradient norm of the deformation field. DARE further introduces a Jacobian-determinant-based folding prevention mechanism that explicitly penalizes regions with negative Jacobian values. By integrating gradient-aware modulation with biophysically grounded constraints, DARE ensures diffeomorphic and anatomically plausible transformations. Extensive experiments on multimodal medical images demonstrate that DARE significantly improves registration accuracy (2.1% Dice score gain) and anatomical consistency while effectively eliminating folding artifacts—outperforming state-of-the-art unsupervised registration methods.

Technology Category

• Computer Vision: Medical and Biological Imaging
• Machine Learning: Learning with Manifolds
• Intelligent Robots: Manipulation

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Deformable medical image registration is a fundamental task in medical image analysis. While deep learning-based methods have demonstrated superior accuracy and computational efficiency compared to traditional techniques, they often overlook the critical role of regularization in ensuring robustness and anatomical plausibility. We propose DARE (Deformable Adaptive Regularization Estimator), a novel registration framework that dynamically adjusts elastic regularization based on the gradient norm of the deformation field. Our approach integrates strain and shear energy terms, which are adaptively modulated to balance stability and flexibility. To ensure physically realistic transformations, DARE includes a folding-prevention mechanism that penalizes regions with negative deformation Jacobian. This strategy mitigates non-physical artifacts such as folding, avoids over-smoothing, and improves both registration accuracy and anatomical plausibility

Problem

• Enhancing robustness in learning-based medical image registration
• Dynamically adapting elastic regularization for deformation fields
• Preventing non-physical artifacts while maintaining anatomical plausibility

Innovation

• DARE dynamically adjusts elastic regularization based on deformation gradient
• It integrates adaptively modulated strain and shear energy terms
• Includes folding-prevention mechanism penalizing negative Jacobian regions