Multi-structure bone segmentation in pediatric MR images with combined regularization from shape priors and adversarial network
作者：
Highlights：
• A deep learning multi-structure segmentation method for pediatric bones in MR images.
• We present a training strategy based on shape priors and adversarial regularizers.
• The combined regularization enforces globally consistent and plausible segmentation.
• Our model learns bone-specific features and characteristics shared among bones.
• Comparison with state-of-the-art shows improved performance on two sparse datasets.
摘要
•A deep learning multi-structure segmentation method for pediatric bones in MR images.•We present a training strategy based on shape priors and adversarial regularizers.•The combined regularization enforces globally consistent and plausible segmentation.•Our model learns bone-specific features and characteristics shared among bones.•Comparison with state-of-the-art shows improved performance on two sparse datasets.
论文关键词：Deep learning,Anatomical priors,Adversarial networks,Ankle,Shoulder
论文评审过程：Received 17 November 2021, Revised 13 May 2022, Accepted 10 July 2022, Available online 19 July 2022, Version of Record 30 July 2022.
论文官网地址：https://doi.org/10.1016/j.artmed.2022.102364
