摘要
针对腹部多器官图像分割中器官特征提取不足及边缘模糊导致的分割精度低的问题,提出基于DFCU-Net的腹部多器官图像分割模型。首先,采用双编码器结构协同提取腹部多器官图像的局部和全局特征;其次,采用特征融合模块将局部和全局特征充分融合;最后,在解码器中采用特征重组上采样模块以减少器官边缘特征丢失,进而实现图像分割。实验结果表明,DFCU-Net模型的平均Dice相似系数和Hausdorff距离均达到最佳值,分别为81.43%和25.80mm。与其他模型相比,DFCU-Net模型显著提高了腹部多器官图像分割精度。
Aiming at the problem of low segmentation accuracy caused by insufficient feature extraction of organs and blurred edges in abdominal multi organ image segmentation,an abdominal multi-organ image segmentation model based on DFCU-Net is proposed.Firstly,a dual-encoder structure is adopted to collaboratively extract local and global features of abdominal multi-organ images.Secondly,a feature fusion module is used to fully fuse the local and global features.Finally,a feature recombination up sampling module is used in the decoder to reduce the loss of organ edge features and achieve image segmentation.The experimental results show that the average Dice similarity coefficient and Hausdorff distance of the DFCU-Net model reach their optimal values of 81.43%and 25.80mm,respectively.Compared with other models,the DFCU-Net model significantly improves the accuracy of abdominal multi organ image segmentation.
作者
陈静
汪含丹
吴齐婷
徐磊
Chen Jing;Wang Handan;Wu Qiting;Xu Lei(School of Electrical and Information Engineering,Anhui University of Science and Technology,Huainan,Anhui 232001,China)
出处
《黑龙江工业学院学报(综合版)》
2025年第11期88-92,共5页
Journal of Heilongjiang University of Technology(Comprehensive Edition)
基金
国家自然科学基金项目(项目编号:51874010)
安徽省教育厅高校自然科学研究项目(项目编号:KJ2018A0087)。
关键词
腹部多器官图像分割
双编码器
特征融合
特征重组上采样模块
abdominal multi-organ image segmentation
dual encoder structure
feature fusion
feature recombination up sampling module
