Evaluation of Alignment-Regularity Characteristics in Deformable Image Registration

Evaluating deformable image registration (DIR) is challenging due to the inherent trade-off between achieving high alignment accuracy and maintaining deformation regularity. In this work, we introduce a novel evaluation scheme based on the alignment-regularity characteristic (ARC) to systematically capture and analyze this trade-off. We first introduce the ARC curves, which describe the performance of a given registration algorithm as a spectrum measured by alignment and regularity metrics. We further adopt a HyperNetwork-based approach that learns to continuously interpolate across the full regularization range, accelerating the construction and improving the sample density of ARC curves. We empirically demonstrate our evaluation scheme using representative learning-based deformable image registration methods with various network architectures and transformation models on two public datasets. We present a range of findings not evident from existing evaluation practices and provide general recommendations for model evaluation and selection using our evaluation scheme. All code relevant is made publicly available.

@article{sideri-lampretsa2025_2503.07185,
  title={ Evaluation of Alignment-Regularity Characteristics in Deformable Image Registration },
  author={ Vasiliki Sideri-Lampretsa and Daniel Rueckert and Huaqi Qiu },
  journal={arXiv preprint arXiv:2503.07185},
  year={ 2025 }
}