Skip to main content

Table 4 Comparison of quantitative results of different methods on the Drishti-GS1 dataset

From: RSAP-Net: joint optic disc and cup segmentation with a residual spatial attention path module and MSRCR-PT pre-processing algorithm

Method

Year

OD (mean/std)

OC (mean/std)

F1

BLE(px)

F1

BLE (px)

FCN [20]

2014

0.9321/0.102

8.90/5.74

0.8170/0.103

21.83/15.67

U-Net [21]

2015

0.9600/0.020

7.23/4.51

0.8500/0.100

19.53/13.98

M-Net [18]

2018

0.9590/0.040

7.97/8.29

0.866/0.110

17.05/12.76

Stack-U-Net [44]

2018

0.9700/0.020

6.47/3.51

0.8900/0.090

14.39/7.18

POSAL [34]

2019

0.9650/–

–/–

0.8580/–

–/–

CE-Net [32]

2019

0.9688/0.003

5.04/3.69

0.8699/0.117

16.06/13.11

JointRCNN [38]

2020

0.9640/–

–/–

0.8640/–

–/–

BGA-Net [45]

2021

0.9750/0.019

7.01/3.53

0.8980/0.08

14.37/9.29

DCGAN [46]

2022

0.9746/0.017

7.35/4.83

0.8631/0.113

18.69/13.48

Proposed method

–

0.9752/0.012

6.33/3.30

0.9012/0.08

11.97/6.12