GLIDE-Reg: Global-to-Local Deformable Registration Using Co-Optimized Foundation and Handcrafted Features

2026-02-27

AI Summary

Medical image registration faces challenges in robustness and generalization under cross-scale and cross-anatomical-region scenarios. This work proposes a joint optimization framework that, for the first time, integrates a learnable dimensionality reduction module with a compressed vision foundation model (VFM) and incorporates MIND local descriptors to enable synergistic alignment of global semantic and local structural information. The method achieves average Dice similarity coefficients (DSC) of 0.859, 0.862, and 0.901 on the Lung250M, NLST, and UCLA5DCT datasets, respectively, outperforming state-of-the-art approaches such as DEEDS. Furthermore, it attains a localization accuracy of 1.11 mm in pulmonary nodule center detection, demonstrating its practical utility in early lung cancer diagnosis.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Search and Optimization: Learning to Search

Application Category

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

Abstract

Deformable registration is crucial in medical imaging. Several existing applications include lesion tracking, probabilistic atlas generation, and treatment response evaluation. However, current methods often lack robustness and generalizability across two key factors: spatial resolution and differences in anatomical coverage. We jointly optimize a registration field and a learnable dimensionality reduction module so that compressed VFM embeddings remain registration-relevant, and fuse these global semantic cues with MIND local descriptors. GLIDE-Reg achieves average dice similarity coefficients (DSC) across 6 anatomical structures of 0.859, 0.862, and 0.901 in two public cohorts (Lung250M and NLST) and one institution cohort (UCLA5DCT), and outperforms the state-of-the-art DEEDS (0.834, 0.858, 0.900) with relative improvements of 3.0%, 0.5%, and 0.1%. For target registration errors, GLIDE-Reg achieves 1.58 mm on Lung250M landmarks (compared to 1.25 mm on corrField and 1.91 mm on DEEDS) and 1.11 mm on NLST nodule centers (compared to 1.11 mm on DEEDS). The substantiated performance on the nodule centers also demonstrates its robustness across challenging downstream tasks, such as nodule tracking, which is an essential prior step for early-stage lung cancer diagnosis.

Problem

• deformable registration
• robustness
• generalizability
• medical imaging
• anatomical coverage

Innovation

• deformable registration
• foundation model
• feature fusion
• dimensionality reduction
• medical image analysis