Skip to main content

Table 2 Mean±STD of our GAN method and other traditional methods are summarized. The GTF, FPDF, and GFDF will be excluded from the quantitative comparisons as they did not transfer MRI soft-tissue contrast, add a noticeable amount of spatial distortion, and did not transfer CT bone information, respectively. Bold indicates the best results. Underline indicate a better result than ours that was excluded because it did not satisfy the fusion criteria

From: MedFusionGAN: multimodal medical image fusion using an unsupervised deep generative adversarial network

Method

ENT

STD

PSNR

\(Q^{XY/F}\)

MG

SF

NCC

MI

SSIM

GTF

5.25\(\pm\)0.42

0.33\(\pm\)0.03

4.15\(\pm\)3.91

0.6\(\pm\)0.06

\(\underline{0.21\!\pm \!0.05}\)*

\(\underline{0.64\!\pm \!0.1}\)*

0.82\(\pm\)0.12

\(\underline{0.39\!\pm \!0.21}\)*

0.29\(\pm\)0.37

DDCT-PCA

5.13\(\pm\)0.3*

0.37\(\pm\)0.05

5.98\(\pm\)1.65

0.6\(\pm\)0.06

0.18\(\pm\)0.04

0.53\(\pm\)0.08*

0.85\(\pm\)0.11

0.42\(\pm\)0.2*

0.33\(\pm\)0.16

FPDE

4.77\(\pm\)0.34

0.27\(\pm\)0.05

8.46\(\pm\)1.65

0.37\(\pm\)0.10

\(\underline{0.20\!\pm \!0.04}\)*

0.77\(\pm\)0.14

0.3\(\pm\)0.28

0.58\(\pm\)0.18

0.25\(\pm\)0.26

HMSD

5.58\(\varvec{\pm }\)0.23

0.27\(\pm\)0.04

5.75\(\pm\)1.86

0.6\(\pm\)0.07

0.21\(\varvec{\pm }\)0.05*

0.61\(\pm\)0.10

0.82\(\pm\)0.12

0.35\(\pm\)0.22

0.32\(\pm\) 0.17

GFDF

2.07 ± 0.51

0.23 \(\pm\) 0.03

16.68 \(\pm\) 1.64

0.46 \(\pm\) 0.04

0.08 \(\pm\) 0.02

0.37 \(\pm\) 0.06

0.78 \(\pm\) 0.22

\(\underline{0.53\!\pm \!0.43}\)*

\(\underline{0.66\!\pm \!0.33}\)*

IVF

2.74± 0.57

0.22± 0.03

12.69 ± 2.79

0.54 ± 0.04

0.11 ± 0.03

0.49 ± 0.09

0.83 ± 0.06

0.4 ± 0.36*

0.13 ± 0.05

MEF

2.06± 0.48

0.25± 0.03

20.38± 1.86

0.61± 0.06

0.06± 0.02

0.33± 0.06

0.82± 0.04

0.57\(\varvec{\pm }\)0.04

0.81\(\varvec{\pm }\)0.03

Ours

5.2\(\pm\)0.38

0.44\(\varvec{\pm }\)0.05

23.02\(\varvec{\pm }\)3.5

0.64\(\varvec{\pm }\)0.1

0.20\(\varvec{\pm }\)0.05

0.67\(\varvec{\pm }\)0.14

0.91\(\varvec{\pm }\)0.04

0.42\(\pm\)0.29

0.62\(\pm\)0.22

  1. * is not statistically different (p-value \(> 0.05\)) from our proposed MedFusionGAN method.
  2. Abbreviations: ENT entropy, STD standard deviation, PSNR peak signal-to-noise ratio, MG mean gradient, SF spatial frequency, NCC normalized cross-correlation, MI mutual information, SSIM structural similarity index