Dynamic Stream Network for Combinatorial Explosion Problem in Deformable Medical Image Registration

TL;DR

Deformable medical image registration suffers from combinatorial explosion when modeling feature relationships across dual inputs. The authors introduce DySNet, which adaptively modulates both receptive fields (AdSB) and attention weights (DySA) in a bidirectional framework to suppress interfering feature combinations and improve relationship discovery. Across three tasks (3D cardiac, 3D brain, and 2D brain) and multiple baselines, DySNet achieves higher Dice scores and smoother deformation fields, indicating strong generalization and robustness. The work provides modular dynamic components and demonstrates substantial potential for broader application in medical image analysis, with code to be released publicly.

Abstract

Combinatorial explosion problem caused by dual inputs presents a critical challenge in Deformable Medical Image Registration (DMIR). Since DMIR processes two images simultaneously as input, the combination relationships between features has grown exponentially, ultimately the model considers more interfering features during the feature modeling process. Introducing dynamics in the receptive fields and weights of the network enable the model to eliminate the interfering features combination and model the potential feature combination relationships. In this paper, we propose the Dynamic Stream Network (DySNet), which enables the receptive fields and weights to be dynamically adjusted. This ultimately enables the model to ignore interfering feature combinations and model the potential feature relationships. With two key innovations: 1) Adaptive Stream Basin (AdSB) module dynamically adjusts the shape of the receptive field, thereby enabling the model to focus on the feature relationships with greater correlation. 2) Dynamic Stream Attention (DySA) mechanism generates dynamic weights to search for more valuable feature relationships. Extensive experiments have shown that DySNet consistently outperforms the most advanced DMIR methods, highlighting its outstanding generalization ability. Our code will be released on the website: https://github.com/ShaochenBi/DySNet.

Figures (10)

• Figure 1: Combinatorial explosion problem in DMIR: (a) dual inputs cause a combinatorial explosion of features. (b) static receptive fields contains a large number of interfering features. (c) static weights ignore potential accuracy features. (d) ideal situation where dynamic receptive fields and weights adaptively model the feature relationships.
• Figure 2: Motivation: (a) shows that static receptive fields and static weights restrict learning ability by limiting the search space and direction. (b) illustrates dynamic receptive fields enable the adaptability of relationship search space. (c) demonstrates that dynamic weights allow for precise localization of feature relationship, enhancing matching accuracy by adjusting the search space and direction dynamically in the embedding space.
• Figure 3: Overall architecture of our DySNet. a) Dynamic stream block generates queries, keys, and values according to fixed features and moving features, and combines the AdSB and the DySA to respectively construct dynamic receptive fields and dynamic weights, ultimately forming attention features. b) The AdSB learns the dynamic receptive fields. c) The DySA calculates dynamic weights of feature modeling. d) the general network architecture of DySNet.
• Figure 4: This figure presents a qualitative comparison of various registration methods between dynamic images and fixed images. It includes our proposed DySNet, as well as mainstream registration networks. The results show that our method outperforms others in maintaining details and structural consistency and it also significantly improves upon methods with the same network structure (ModeT and Xmorpher). “*" mean the values are invalid because the methods registration failed.
• Figure 5: a) Ablation results of our DySNet-X on the 2D Brain MRI set. Add the bidirectional registration (bi), DySA and AdSB modules in sequence. b) Hyper-parameter ablation of kernel sizes.