Encoder-Only Image Registration
- 6 January 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 36 (5) , 7035-7049
Abstract
Learning-based techniques have significantly improved the accuracy and speed of deformable image registration. However, challenges such as reducing computational complexity and handling large deformations persist. To address these challenges, we analyze how convolutional neural networks (ConvNets) influence registration performance using the Horn-Schunck optical flow equation. Supported by prior studies and our empirical experiments, we observe that ConvNets play two key roles in registration: linearizing local intensities and harmonizing global contrast variations. Guided by these insights, we propose the Encoder-Only Image Registration (EOIR) framework comprising five modifications to existing approaches, to achieve a better accuracy-efficiency trade-off. EOIR separates feature learning from flow estimation, employing only a 3-layer ConvNet for feature extraction and a set of 3-layer flow estimators to construct a Laplacian feature pyramid, progressively composing diffeomorphic deformations under a large-deformation model. Results on six datasets across different modalities and anatomical regions demonstrate EOIR’s effectiveness, achieving superior accuracy-efficiency and accuracy-smoothness trade-offs. With comparable accuracy, EOIR provides better efficiency and smoothness, and vice versa. The source code of EOIR is available on Github.Keywords
All Related Versions
Funding Information
- National Natural Science Foundation of China (U22B2050)
- National Natural Science Foundation of China (62425305)
- National Natural Science Foundation of China (62503161)
- Natural Science Foundation of Hunan Province (2025JJ60389)
- Congressionally Directed Medical Research Programs (HT9425-25-1-0716)
This publication has 80 references indexed in Scilit:
- ReCoNet: Recurrent Correction Network for Fast and Efficient Multi-modality Image Fusion2022
- Fast 3D Registration with Accurate Optimisation and Little Learning for Learn2Reg 20212022
- A Deep Discontinuity-Preserving Image Registration Network2021
- HyperMorph: Amortized Hyperparameter Learning for Image Registration2021
- Large Deformation Image Registration with Anatomy-Aware Laplacian Pyramid Networks2021
- Large Deformation Diffeomorphic Image Registration with Laplacian Pyramid Networks2020
- Unsupervised Learning for Fast Probabilistic Diffeomorphic Registration2018
- A longitudinal four‐dimensional computed tomography and cone beam computed tomography dataset for image‐guided radiation therapy research in lung cancerMedical Physics, 2016
- Quantitative susceptibility mapping (QSM): Decoding MRI data for a tissue magnetic biomarkerMagnetic Resonance in Medicine, 2014
- A Log-Euclidean Framework for Statistics on Diffeomorphisms2006