Table 3 Cross-domain generalization: Training on LiTS (CT) and testing on 3DIRCADb (CT) without fine-tuning

From: Prompt-mamba filtering networks for accurate hepatocellular carcinoma lesion segmentation in abdominal CT

Method

Dice (%)

HD95 (mm)

Recall (%)

Prec. (%)

U-Net

67.8

14.2

70.3

68.1

Attention U-Net

69.1

13.5

72.0

69.4

U-Net++

70.5

12.7

73.4

70.8

U-Net3+

71.2

12.0

74.1

71.3

TransUNet

72.6

11.1

75.2

73.0

Swin-UNet

74.5

9.3

77.0

74.2

MISSFormer

73.9

9.0

76.5

73.6

HiFormer

74.7

8.8

77.2

74.9

MERIT

76.1

8.6

78.0

75.6

MedSAM

76.4

8.5

78.1

75.9

MedSegDiff

75.2

8.9

77.0

74.7

Prompt-Mamba-AF

79.2

7.4

82.5

77.6

  1. The bold means the best results in the table.
Back to article page