VoxelOpt: Voxel-Adaptive Message Passing for Discrete Optimization in Deformable Abdominal CT Registration

2025-06-24

AI Summary

Existing learning-based deformable image registration (DIR) methods suffer from limited accuracy under small-sample, large-deformation, or unsupervised settings, whereas traditional iterative approaches achieve high accuracy at prohibitive computational cost. To address this trade-off, we propose VoxelOpt—a hybrid learning-and-discrete-optimization framework for abdominal CT registration. VoxelOpt constructs a 27-neighborhood cost volume across a multi-scale image pyramid via voxel-adaptive message passing; leverages features from a pre-trained segmentation model to extract robust anatomical priors—bypassing hand-crafted design and contrastive learning; and incorporates displacement-entropy-driven local cost analysis to enhance deformation modeling. Experiments demonstrate that VoxelOpt surpasses state-of-the-art iterative methods in accuracy while matching the inference speed of leading supervised learning-based approaches. Crucially, it achieves superior accuracy–efficiency balance in label-free scenarios, advancing practical deployment of DIR in clinical workflows.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Mixed Discrete/Continuous Search
• Constraint Satisfaction and Optimization: Mixed Discrete/Continuous Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Recent developments in neural networks have improved deformable image registration (DIR) by amortizing iterative optimization, enabling fast and accurate DIR results. However, learning-based methods often face challenges with limited training data, large deformations, and tend to underperform compared to iterative approaches when label supervision is unavailable. While iterative methods can achieve higher accuracy in such scenarios, they are considerably slower than learning-based methods. To address these limitations, we propose VoxelOpt, a discrete optimization-based DIR framework that combines the strengths of learning-based and iterative methods to achieve a better balance between registration accuracy and runtime. VoxelOpt uses displacement entropy from local cost volumes to measure displacement signal strength at each voxel, which differs from earlier approaches in three key aspects. First, it introduces voxel-wise adaptive message passing, where voxels with lower entropy receives less influence from their neighbors. Second, it employs a multi-level image pyramid with 27-neighbor cost volumes at each level, avoiding exponential complexity growth. Third, it replaces hand-crafted features or contrastive learning with a pretrained foundational segmentation model for feature extraction. In abdominal CT registration, these changes allow VoxelOpt to outperform leading iterative in both efficiency and accuracy, while matching state-of-the-art learning-based methods trained with label supervision. The source code will be available at https://github.com/tinymilky/VoxelOpt

Problem

• Balances accuracy and speed in deformable image registration
• Addresses large deformations and limited training data challenges
• Improves voxel-wise adaptive message passing for better precision

Innovation

• Voxel-wise adaptive message passing for optimization
• Multi-level pyramid with 27-neighbor cost volumes
• Pretrained segmentation model for feature extraction