Feature-centric registration of large deformed images using transformers and correlation distance

• Published in: Computers in biology and medicine 2025-01, Vol.184, p.109356, Article 109356
• Main Authors: Kim, Heeyeon, Lee, Minkyung, Kim, Bohyoung, Shin, Yeong-Gil, Chung, Minyoung
• Format: Article
• Language: English
• Subjects: Algorithms; Coarse-to-fine transformers; Correlation analysis; Correlation-based distance metric; Cross and self attentions; Deformable image registration; Deformation; Feature extraction; Feature-centric registration; Fields (mathematics); Formability; Humans; Image Processing, Computer-Assisted - methods; Image registration; Magnetic Resonance Imaging - methods; Medical imaging; Receptive field; Registration; Robustness; Similarity coefficient method; Tomography, X-Ray Computed - methods
• ISSN: 0010-4825; 1879-0534; 1879-0534
• DOI: 10.1016/j.compbiomed.2024.109356

In deformable medical image registration, both a robust backbone registration network and a suitable similarity metric are essential. This paper introduces a robust registration network combined with a feature-based loss function, specifically designed to handle large deformations and address the challenge of the absence of ground truth data. Tackling large deformations typically requires either expanding the receptive field or breaking down extensive deformations into smaller, more manageable ones. We address this challenge through two key network components: the coarse-to-fine estimation of the target displacement vector field (DVF) and the integration of the Transformer’s feature attention mechanism. To further enhance registration performance, we propose a novel feature correlation-based distance metric that leverages the symmetric properties of the correlation matrix to efficiently exploit feature correlations. Additionally, by utilizing the features extracted directly from the registration network, we eliminate the need for additional feature extraction networks. Experimental results demonstrate that our feature correlation-based loss function is particularly effective in achieving accurate registration in the absence of ground truth data. Our method has proven successful in both mono-modality abdomen CT registration and brain MRI atlas registration, leading to improvements in Dice similarity coefficient and other evaluation metrics. •A novel registration loss for large deformations in an unsupervised setting.•Utilizes Transformer attention and a coarse-to-fine approach for large deformations.•Employs a feature correlation distance loss for effective unsupervised learning.•Unlike existing feature losses, it avoids training an additional feature extractor.