From: Automatic segmentation of large-scale CT image datasets for detailed body composition analysis
Target | Architecture (performance) |
---|---|
Liver | 2D EfficientNet based UNET [27] (0.960); PADLL [28] (0.965); *UNET++ (0.994) |
Spleen | 2D EfficientNet based UNET [27] (0.950); *UNET++ (0.993) |
VAT | UNET [11] (0.940); FCN-based Segmentation [12] (0.970); UNET (pretrain VGG-16 encoder) [14] (0.960); 3D(RGA-UNET) and Standard 3D UNET [29] (0.90) Automatic segmentation method [30] (0.955); CNN (encoder and decoder) [31] (0.970); FatSegNet [32] (0.990); UNET [33] (0.968); Fast graph-based algorithm [34] (0.996); UNET [35] (0.997); UNET [36] (0.9746); Fully automatic segmentation algorithm [37] (0.920); UNET [38] (0.970); *UNET++ (0.973) |
SAT | UNET [11] (0.940); FCN-based Segmentation [12] (0.970); Automatic segmentation method [30] (0.972); CNN (encoder and decoder) [31] (0.980); FatSegNet [32] (0.990); UNET [33] (0.968); Fast graph-based algorithm [34] (0.996); UNET [35] (0.998); UNET [36] (0.943); UNET [38] (0.960); UNET [39] (0.970); UNET [40] (0.976); *Ghost-UNET++ (0.994) |
DSAT/SSAT | UNET (pretrain VGG-16 encoder) [14] (0.909/0.960); 3D(RGA-UNET) and Standard 3D UNET [29] (0.880/0.920); Fully automatic segmentation algorithm [37] (0.820/0.880); *UNET++ (0.972/0.968) |
Spine bone marrow | UNET [39] (0.920); *UNET++ (0.993) |
Skeleton muscle | FCN-based segmentation [12] (0.970); automatic segmentation method [30] (0.952); UNET [39] (0.950); *UNET++ (0.988) |
IMAT | CNN (encoder and decoder) [31] (0.830); UNET [39] (0.910); *UNET++ (0.927) |