Table 2 Comparison experiment of semantic segmentation methods.

From: A full-scale lung image segmentation algorithm based on hybrid skip connection and attention mechanism

model

Precision

Recall

PA

Dice

mIoU

FCN15

0.958 ± 0.017

0.947 ± 0.01

0.968 ± 0.02

0.943 ± 0.014

0.927 ± 0.02

U-net6

0.949 ± 0.023

0.936 ± 0.021

0.966 ± 0.018

0.972 ± 0.011

0.919 ± 0.021

UNet +  +17

0.957 ± 0.018

0.933 ± 0.008

0.964 ± 0.016

0.960 ± 0.013

0.927 ± 0.023

AAF-U-Net16

0.956 ± 0.014

0.929 ± 0.022

0.968 ± 0.014

0.961 ± 0.017

0.931 ± 0.025

PSPNet18

0.961 ± 0.011

0.944 ± 0.01

0.971 ± 0.01

0.963 ± 0.013

0.939 ± 0.015

DeepLab v310

0.958 ± 0.02

0.951 ± 0.012

0.974 ± 0.015

0.951 ± 0.019

0.972 ± 0.015

HAFS (Ours)

0.966 ± 0.01

0.958 ± 0.014

0.977 ± 0.016

0.968 ± 0.021

0.986 ± 0.01

Back to article page