Skip to main content

Table 4 Quantitative assessment of fat and water predictions compared with the original data, using the data of 65 subjects with manual annotation data

From: Artifact-free fat-water separation in Dixon MRI using deep learning

Model

Water

Fat

SIS

SSIM

PSNR (dB)

SSIM

PSNR (dB)

Spleen

Kidney

Liver

Iliopsoas

\(IP\rightarrow \hat{F},\hat{W}\)

0.927 \({\pm }\) 0.011

24.83 \({\pm }\) 1.00

0.950 \({\pm }\) 0.011

25.25 \({\pm }\) 1.12

0.858

0.802

0.965

0.935

\(IP,OP\rightarrow \hat{F},\hat{W}\)

0.967 \({\pm }\) 0.008

29.47 \({\pm }\) 1.57

0.976 \({\pm }\) 0.009

29.73 \({\pm }\) 1.67

0.926

0.934

0.988

0.990

\(IP,OP\rightarrow \hat{F},\hat{W}\)

0.941 \({\pm }\) 0.009

26.50 \({\pm }\) 1.42

0.959 \({\pm }\) 0.010

27.01 \({\pm }\) 1.56

0.895

0.898

0.965

0.959

(Dixon generator loss)

  1. In addition to PSNR and SSIM, we computed the semantic interpretability scores (SIS) for the spleen, kidney, liver and iliopsoas muscle segmentations