Table 6 Quantitative comparison of the image quality analysis of ten (10) randomly selected synthesized images with architectural distortion (AD) for metrics ranging in the categories of reference-based, nonreference-based, and feature-based.

From: A generative adversarial network for synthetization of regions of interest based on digital mammograms

Images

PSNR

SSIM

DSSIM

MSE

FSIM

BRISQUE

PQUE

NIQE

GS

FID

1

27.97

0.04

0.48

103.73

0.90

27.05

22.65

25.81

0.00

343,130.76

2

28.00

0.04

0.48

103.07

0.88

17.79

23.17

37.07

0.00

321,157.06

3

27.94

0.04

0.48

104.41

0.89

29.11

23.04

30.40

0.00

413,143.07

4

27.83

0.03

0.48

107.18

0.89

15.33

22.70

28.10

0.00

429,359.88

5

27.98

0.04

0.48

103.47

0.89

28.32

23.67

25.31

0.00

330,390.83

6

28.00

0.05

0.48

103.09

0.88

16.62

23.71

27.50

0.00

339,042.88

7

27.93

0.05

0.47

104.61

0.87

34.25

24.23

22.30

0.00

425,196.46

8

27.99

0.04

0.48

103.22

0.87

34.27

23.26

27.56

0.00

308,565.14

9

28.02

0.06

0.47

102.61

0.90

18.04

23.01

27.48

0.00

260,963.64

10

28.04

0.05

0.47

102.22

0.87

37.61

24.91

24.45

0.00

335,339.00

Average

27.97

0.04

0.48

103.76

0.88

25.84

23.44

27.60

0.00

350,628.87