多尺度残差学习结合Dilformer的双流医学图像配准网络

彭静,闫佳荣,刘佳英,魏子易,白珊,邓亚红

Multi-scale residual learning combined with Dilformer for dual-stream medical image registration network

Jing PENG,Jiarong YAN,Jiaying LIU,Ziyi WEI,Shan BAI,Yahong DENG

表 4 在LPBA40数据集上的模型泛化性验证数据

Tab.4 Generalization validation data of model on LPBA40 dataset

模型 DSC/% HD95 $ {R}_{{\mathrm{Jac}}} $ t/s SyN 62.7±1.2 7.59±1.62 <0.001 35.8 LDDMM 61.3±1.5 7.61±1.53 <0.001 32.6 VoxelMorph 65.7±2.9 7.65±1.47 0.607 0.20 CycleMorph 67.1±2.9 7.58±1.45 0.497 0.22 LKU-Net 70.3±1.5 7.39±1.57 0.203 0.29 ViT-V-Net 67.0±2.9 7.57±1.72 0.207 0.35 TransMorph 69.4±2.1 7.43±1.85 0.161 0.36 TransMatch 70.3±1.6 7.49±1.44 0.183 0.35 PIVit 71.5±1.5 6.51±1.56 0.022 0.26 RDP-Net 71.6±1.7 6.43±1.83 <0.001 0.28 MSRD-Net 72.9±1.5 6.32±1.71 0.117 0.27