Table 6

Quantitative comparisons between MSAN and other methods in terms of PSNR and SSIM with the proposed 2D ultrasound imaging dataset. The first and second best performances are denoted in red and blue, respectively.

PSNR(dB)/SSIM
MethodsThe number of raw images for averagingAverage value
124816
Traditional methodNoised20.67 / 0.298021.00 / 0.391723.00 / 0.477023.67 / 0.555923.75 / 0.661022.42 / 0.4767
NLM21.22 / 0.642823.03 / 0.717925.99 / 0.732827.88 / 0.744829.32 / 0.798525.49 / 0.7274
BM3D21.38 / 0.694124.47 / 0.714726.06 / 0.749928.43 / 0.777428.78 / 0.809325.82 / 0.7491
KSVD21.50 / 0.663524.02 / 0.697124.94 / 0.758826.90 / 0.789830.15 / 0.789825.50 / 0.7398
EPLL21.31 / 0.685223.76 / 0.731725.31 / 0.746727.79 / 0.775929.83 / 0.813825.60 / 0.7507
WNNM22.19 / 0.671823.55 / 0.729325.23 / 0.746827.27 / 0.756029.67 / 0.763425.58 / 0.7335
PURE-LET22.42 / 0.678323.20 / 0.734425.95 / 0.777528.30 / 0.751129.54 / 0.786325.88 / 0.7455
Early DLDnCNN26.08 / 0.749927.89 / 0.763229.36 / 0.802130.67 / 0.784330.89 / 0.844428.98 / 0.7888
IRCNN25.86 / 0.768828.07 / 0.788328.60 / 0.790030.63 / 0.834930.74 / 0.808028.78 / 0.8000
MemNet25.23 / 0.738926.89 / 0.813428.33 / 0.793030.17 / 0.780231.77 / 0.828728.48 / 0.7948
SotA DLNoise2Noise26.80 / 0.754127.48 / 0.774829.16 / 0.774829.90 / 0.779932.35 / 0.801729.14 / 0.7771
MWCNN26.34 / 0.744827.48 / 0.781528.75 / 0.765430.61 / 0.804930.62 / 0.779428.76 / 0.7752
RIDNet25.86 / 0.726326.79 / 0.790828.90 / 0.793230.74 / 0.790930.73 / 0.807928.60 / 0.7818
DPDN26.87 / 0.734027.01 / 0.760828.96 / 0.827931.10 / 0.815532.11 / 0.824729.21 / 0.7926
WF-UNet25.40 / 0.628327.98 / 0.703429.26 / 0.730629.80 / 0.748730.22 / 0.739428.53 / 0.7101
MSAN (Ours)27.88 / 0.782030.24 / 0.829030.80 / 0.833832.76 / 0.840833.25 / 0.833030.98 / 0.8233
[ViewLarge]
This Feature Is Available To Subscribers Only

or Create an Account

Close Modal
Close Modal