Skip to main content

Table 1 Experiment results of skin segmentation for the ISIC2018 dataset

From: iU-Net: a hybrid structured network with a novel feature fusion approach for medical image segmentation

Family

Methods

Year

Dice(%)

IoU(%)

Precision(%)

Recall(%)

CNN

UNet [1]\(^{\ast}\)  

2015

79.89±5.09

71.02±6.69

84.04±4.38

82.01±4.42

 

Atten_UNet [3]\(^{\ast}\)  

2018

88.15±8.96

81.21±7.23

85.25±5.85

84.98±5.53

 

Channel_Unet [38]

2019

84.82

75.92

94.01

81.04

 

ResUNet [6]\(^{\ast}\)  

2019

79.15

70.15

82.43

84.77

 

CENet [4]\(^{\ast}\)  

2019

89.53±2.81

82.60±4.53

92.81±4.08

86.76±4.95

 

CA-Net [5]\(^{\ast}\)  

2020

90.05±2.43

-

-

-

 

PraNet [39]

2021

87.46

80.23

91.28

87.59

 

AS-Net [40]

2022

89.55

83.09

-

93.06

 

Ms RED [41]\(^{\ast}\)  

2022

87.69±0.53

82.37±0.62

91.87±0.32

88.16±0.58

MLP

UNeXt [42]\(^{\ast}\)  

2022

89.21±0.79

82.1±1.26

-

-

Transformer

SwinUnet [26]\(^{\ast}\)  

2021

88.87

81.67

94.70

86.07

 

MedT [43]

2021

87.35±0.18

79.54±0.26

-

-

 

ViT-B_16 [16]\(^{\ast}\)  

2021

87.54

80.73

94.20

87.21

 

TransUNet(ViT) [10]\(^{\ast}\)  

2021

88.91

81.67

93.05

87.74

 

TransUNet(R50) [10]\(^{\ast}\)  

2021

89.71

82.79

94.19

88.21

 

FAT-Net [44]

2022

88.9

81.6

-

-

 

iU-Net(Ours)

2022

90.12

83.06

94.37

88.07

  1. Model results with “*” are reproduced from the published source code. Those with “-” indicate that the corresponding metric results are not provided