Table 4 Comparative benchmark of SOTA 2D segmentation networks on CirrMRI600+  → T2W 2D.

From: Large Scale MRI Collection and Segmentation of Cirrhotic Liver

Method

mIoU

Dice

HD95

Precision

Recall

ASSD

UNet

0.6772

0.6900

38.22

0.7112

0.7592

10.11

AttentionUNet

0.7089

0.7288

36.19

0.7377

0.7689

9.28

nnUnet-2D

0.7229

0.7418

34.56

0.7662

0.7999

8.78

Trasunet

0.7219

0.7457

31.11

0.7447

0.7812

8.66

Synergynet

0.7383

0.7592

30.94

0.7882

0.8222

7.55

MedSegDiff

0.7489

0.7667

30.89

0.7789

0.8192

7.34

  1. 1Bold shows the best performance, and Italic shows the second-best performance.
Back to article page