This paper proposes an unsupervised cross-modality adaptation method for medical image segmentation using a deep neural network. The method achieves bidirectional adaptation between two modalities through a synergistic image and feature alignment approach. Experiments show that the proposed method outperforms existing state-of-the-art unsupervised methods. Moreover, the proposed method has the potential to significantly improve the performance of medical image segmentation tasks in various clinical applications.
暂无评论