文献类型：期刊文献

英文题名：Hybrid transformer and convolution iteratively optimized pyramid network for brain large deformation image registration

作者：Cui, Xinxin[1];Zhou, Yuee[1];Wei, Caihong[2];Suo, Guodong[1];Jin, Fengqing[1];Yang, Jianlan[1,2]

第一作者：Cui, Xinxin

通信作者：Yang, JL[1];Yang, JL[2]

机构：[1]Gansu Univ Tradit Chinese Med, Sch Med Informat Engn, Lanzhou 730000, Gansu, Peoples R China;[2]Fujian Univ Tradit Chinese Med, Quanzhou Orthoped Traumatol Hosp, Quanzhou 362000, Fujian, Peoples R China

第一机构：甘肃中医药大学

通信机构：[1]corresponding author), Gansu Univ Tradit Chinese Med, Sch Med Informat Engn, Lanzhou 730000, Gansu, Peoples R China;[2]corresponding author), Fujian Univ Tradit Chinese Med, Quanzhou Orthoped Traumatol Hosp, Quanzhou 362000, Fujian, Peoples R China.|[10735]甘肃中医药大学;

年份：2025

卷号：15

期号：1

外文期刊名：SCIENTIFIC REPORTS

收录：;Scopus(收录号：2-s2.0-105004289004);WOS:【SCI-EXPANDED(收录号:WOS:001482953300032)】；

基金：The authors thank the School of Medical Information Engineering at Gansu University of Traditional Chinese Medicine and Quanzhou Orthopedic Hospital for their support of this study.

语种：英文

外文关键词：Enhanced pyramid encoder; Convolution iterative optimization; Transformer; Brain MRI

摘要：In recent years, the pyramid-based encoder-decoder network architecture has become a popular solution to the problem of large deformation image registration due to its excellent multi-scale deformation field prediction ability. However, there are two main limitations in existing research: one is that it over-focuses on the fusion of multi-layer deformation sub-fields on the decoding path, while ignoring the impact of feature encoders on network performance; the other is the lack of specialized design for the characteristics of feature maps at different scales. To this end, we propose an innovative hybrid Transformer and convolution iteratively optimized pyramid network for large deformation brain image registration. Specifically, four encoder variants are designed to study the impact of different structures on the performance of the pyramid registration network. Secondly, the Swin-Transformer module is combined with the convolution iterative strategy, and each layer of the decoder is carefully designed according to the semantic information characteristics of different decoding layers. Extensive experimental results on three public brain magnetic resonance imaging datasets show that our method has the highest registration accuracy compared with 9 cutting-edge registration methods, which fully verifies the effectiveness and application potential of our model design.