Table 3 Image quality parameters: mean SNR and CNR with standard deviation.

From: Deep learning reconstruction enhances tophus detection in a dual-energy CT phantom study

 

Raster phantom

Bio-phantom

 

SNR (80 kVp/60 keV)

CNR (80 kVp/60 keV)

SNR (80 kVp/60 keV)

CNR (80 kVp/60 keV)

FBP

11.90 ± 0.77

11.38 ± 0.71

36.53 ± 13,93

16.31 ± 5.83

IR

12.26 ± 0.41

11.69 ± 0.36

55.39 ± 20,57

24.86 ± 9.09

DLR mild

63.52 ± 24.70

59.99 ± 23.66

53.58 ± 16.98

28.42 ± 7.69

DLR strong

112.80 ± 35.98

107.20 ± 34.30

84.03 ± 20.65

43.80 ± 9.98

P

< 0.0001

< 0.0001

< 0.0001

< 0.0001

  1. Mean SNR (signal-to-noise ratio) and CNR (contrast-to-noise ratio) showed significant differences between the four image reconstruction methods for both phantoms.
  2. FBP, filtered back projection; IR, iterative reconstruction (AIDR strong); DLR mild, deep learning reconstruction in mild version; DLR strong, deep learning reconstruction in strong version.
Back to article page