Longitudinally consistent registration and parcellation of cortical surfaces using semi-supervised learning

Fenqiang Zhao, Zhengwang Wu, Li Wang, Weili Lin, Gang Li

Abstract: Highlight•A novel semi-supervised learning-based framework for longitudinally consistent registration and parcellation of cortical surfaces is proposed.•Our method exploits the reciprocal relationship between surface registration and parcellation, enabling more meaningful spatial feature representations.•Our method incorporates a novel longitudinal consistency loss to extract more consistent temporal features.•Experiments demonstrate that our method is more suitable for longitudinal studies compared to popular cross-sectional methods and existing longitudinal strategies.

External IDs:doi:10.1016/j.media.2024.103193