Skip to main content
EJNMMI Physics

Table 2 PSNR and SSIM indices variation in comparison to PET100, considering the slice with all visible spheres from the NEMA IEC phantom experiments (E1 & E2)

From: Clinical and phantom validation of a deep learning based denoising algorithm for F-18-FDG PET images from lower detection counting in comparison with the standard acquisition

 

PET50

PET50 + SP

PET33

PET33 + SP

PSNR (E1)

    

D710

41.61

43.07

36.55

37.42

IQ4

38.86

40.30

37.81

40.26

DMI4

38.53

40.19

33.12

33.88

PSNR (E2)

    

D710

35.60

36.73

35.82

39.29

IQ4

39.27

40.92

34.85

36.98

DMI4

34.71

39.26

30.98

35.04

SSIM (E1)

    

D710

0.982

0.984

0.972

0.976

IQ4

0.979

0.983

0.968

0.977

DMI4

0.983

0.987

0.968

0.973

SSIM (E2)

    

D710

0.980

0.981

0.968

0.975

IQ4

0.980

0.983

0.972

0.978

DMI4

0.976

0.980

0.964

0.972

  1. SubtlePET processing increase the PSNR & SSIM in every instance i.e., PET50 & PET33 images
Back to article page