Author(s)

Xin Wang¹, Fan Zhu^1*, Yaxin Peng¹, Chaomin Shen², Zhen Ye³, Chaozheng Zhou³

¹College of Science, Shanghai University, Shanghai, China.
²School of Computer Science and Technology, East China Normal University, Shanghai, China.
³Shanghai Electric Central Research Institute, Shanghai, China.

The segmentation of unlabeled medical images is troublesome due to the high cost of annotation, and unsupervised domain adaptation is one solution to this. In this paper, an improved unsupervised domain adaptation method was proposed. The proposed method considered both global alignment and category-wise alignment. First, we aligned the appearance of two domains by image transformation. Second, we aligned the output maps of two domains in a global way. Then, we decomposed the semantic prediction map by category, aligning the prediction maps in a category-wise manner. Finally, we evaluated the proposed method on the 2017 Multi-Modality Whole Heart Segmentation Challenge dataset, and obtained 82.1 on the dice similarity coefficient and 4.6 on the average symmetric surface distance, demonstrating the effectiveness of the combination of global alignment and category-wise alignment.

Medical Image Segmentation, Domain Adaptation, Category-Wise Alignment, Cardiac Segmentation

Wang, X. , Zhu, F. , Peng, Y. , Shen, C. , Ye, Z. and Zhou, C. (2021) Semantic Constraint Based Unsupervised Domain Adaptation for Cardiac Segmentation. Advances in Pure Mathematics, 11, 628-643. doi: 10.4236/apm.2021.116041.