传统方法和深度学习用于不同模态心脏医学图像的分割研究进展
常博, 孙灏芸, 高清宇, 王丽嘉

Research Progress on Cardiac Segmentation in Different Modal Medical Images by Traditional Methods and Deep Learning
CHANG Bo, SUN Haoyun, GAO Qingyu, WANG Lijia
表4 基于UCG图像的分割网络总结
Table 4 Summary of different segmentation networks based on UCG images
方法 时间 学习框架 数据集 Dice系数 Hausdorff距离(HD)/mm
左心室
(LV)
右心室
(RV)
心肌
(Myo)
左心室
(LV)
右心室
(RV)
心肌
(Myo)
Multi-domain regularized[78] 2016 Caffe 42894张图像 0.890 - - - - -
CNN[71] 2016 MATLAB 51例临床数据 0.945 - - 1.2648 - -
Deep generative models[130] 2017 TensorFlow 566例临床数据 0.936
Anatomically CNN[68] 2017 - UK Digital Heart、
CETUS、ACDC
0.939 - 0.811 7.89 - 7.12
Shape-guided deformable model
driven by FCN[108]
2018 Keras 69例临床数据 0.86 - - - - -
Recurrent FCN and optical flow[76] 2018 TensorFlow 556例临床病例 0.927 - - - - -
Multi-structure segmentation[75] 2018 - 500例临床数据 0.868 - - 14.3 - -
VoxelAtlasGAN[118] 2018 PyTorch 60例临床病例 0.953 - - 7.26 - -
Automatic biplane[77] 2019 TensorFlow 427例临床病例 0.92 - - - - -
CNN with the active shape model[110] 2019 MATLAB 30例临床数据 0.919 - - 6.38 - -
Time-series information[70] 2020 TensorFlow 211例临床病例 0.695 - - - - -
Beat-to-beat assessment[117] 2020 PyTorch EchoNet-Dynamic 0.92 - - - - -
DPS-Net[114] 2020 PyTorch 10858例临床数据 0.935 - - 5.51 - -
Deep pyramid local[73] 2021 PyTorch CAMUS 0.962 - - 4.6 - -
3D ultrasound evaluation[72] 2022 TensorFlow 26例临床数据 0.82 - - 6.78 - -
Contrastive pretraining[111] 2022 - CAMUS 0.9252 - - - - -
Label-free segmentation[112] 2022 TensorFlow、
Keras
18873例 0.83 - - - - -
Lightweight network[74] 2022 MATLAB 2262例 0.902 - - - - -
GUDU[69] 2023 - CAMUS 0.946 - - 4.7 - -
Knowledge fusion[113] 2023 - EchoNet-Dynamic、
150例临床数据
0.908 - - 6.56 - -