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2 基于CMRI图像的分割网络总结
Table 2 Summary of segmentation networks based on CMRI 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)
Active contour models[87] 2019 TensorFlow ACDC 0.986 0.940 0.969 4.73 5.95 5.42
MSU-Net[42] 2019 TensorFlow ACDC 0.897 0.855 0.836 - - -
3D high resolution[92] 2019 TensorFlow 1912例临床数据 0.8792 - - 3.99 -
Dynamic pixel-wise
weighting-FCN[52]		 2020 TensorFlow MICCAI 2013 - - 0.803 - - -
FCN for left ventricle
segmentation[128]		 2020 - MICCAI2009
33例临床数据		 0.95 - 0.914 - - -
CNN incorporating
domain-specific
constraints[38]		 2020 TensorFlow ACDC 0.959 0.924 0.873 - - -
Combined CNN and
U-net[50]		 2020 PyTorch MICCAI2009 0.951 - - 3.641 - -
Automatic segmentation
and quantification[46]		 2020 PyTorch ACDC、临床数据 0.96 - 0.88 6.31 - 7.11
Fully automatic
segmentation of RV
and LV[88]		 2020 PyTorch ACDC、5570例
临床数据		 0.927 0.873 - - - -
Deep CNN[41] 2020 TensorFlow MICCAI2009 0.961 0.949 0.867 - - -
DMU-net[43] 2020 Keras 71例临床数据 - - - - 4.445 -
Semi-supervised[47] 2021 PyTorch M&Ms 0.909 0.879 0.845 9.42 12.65 11.85
Active contour
models[82]		 2021 - ACDC、LVQuan18 0.890
0.805		 - - 12.247
19.717		 - -
Deep reinforcement
learning[53]		 2021 - ACDC、
Sunnybrook2009		 0.9502
0.9351		 - - - - -
Attention guided U-Net[39] 2021 TensorFlow LVSC - - 0.956 - - 1.456
Dens FCN[36] 2021 TensorFlow 210例临床数据 0.944 0.908 0.851 7.2 7.35 5.9
SegNet[85] 2022 TensorFlow 1354例临床数据 0.878 - - 10.163 - -
FCN[51] 2022 - 150例临床数据 0.930 - - - - -
Cascade approach
structures[49]		 2022 TensorFlow ACDC 0.963 0.900 0.894 8.062 14.660 7.906
DEU-Net2.0[40] 2022 PyTorch ACDC 0.970 0.949 0.904 7.0 12.2 9.0
RNN with Atrous Spatial
pyramid pooling[86]		 2022 PyTorch 56例临床数据 - - 0.8543 - - -
OSFNet[90] 2022 TensorFlow ACDC 0.946 - - 3.976 - -
Deep Atlas network[44] 2023 TensorFlow 71例临床数据 - 0.902 - - 4.358 -