Appearance Transformation Consistency Network for Registering 1H MRI and 129Xe MRI of the lungs

Motivation: 1H MRI and 129Xe MRI offer detailed structural and functional information, crucial for comprehensive lung image analysis. However, aligning them presents challenges due to inherent misregistration. Goal(s): The objective is to develop a method for registration of 1H MRI with 129Xe MRI, correcting motion displacement, which is of great significance for improving the diagnostic accuracy of lung diseases. Approach: We propose a new registration method. The network learns style differences through appearance transformation consistency, and incorporates weakly supervised label loss to capture lung structure. Results: The experimental results indicate that ATC-Net eliminates motion artifacts and achieves precise registration. Impact: Our results will impact radiologists and researchers by enabling precise lung image registration, facilitating new investigations into lung pathologies.