Conditional Deformable Image Registration with Convolutional Neural Network

Recent deep learning-based methods have shown promising results and runtime advantages in deformable image registration. However, analyzing the effects of hyperparameters searching for optimal regularization parameters prove to be too prohibitive methods. This is because it involves training a substantial number separate models with distinct hyperparameter values. In this paper, we propose conditional registration method new self-supervised learning paradigm By features that are correlated hyperparameter, demonstrate solutions arbitrary can captured by single convolutional neural network. addition, smoothness resulting deformation field manipulated strength during inference. Extensive experiments on large-scale brain MRI dataset show our proposed enables precise control without sacrificing advantage or accuracy.