Table 6 Quantitative comparisons between... | Emerald Publishing

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
Methods		The number of raw images for averaging					Average value
Methods		1	2	4	8	16	Average value
Traditional method	Noised	20.67 / 0.2980	21.00 / 0.3917	23.00 / 0.4770	23.67 / 0.5559	23.75 / 0.6610	22.42 / 0.4767
	NLM	21.22 / 0.6428	23.03 / 0.7179	25.99 / 0.7328	27.88 / 0.7448	29.32 / 0.7985	25.49 / 0.7274
	BM3D	21.38 / 0.6941	24.47 / 0.7147	26.06 / 0.7499	28.43 / 0.7774	28.78 / 0.8093	25.82 / 0.7491
	KSVD	21.50 / 0.6635	24.02 / 0.6971	24.94 / 0.7588	26.90 / 0.7898	30.15 / 0.7898	25.50 / 0.7398
	EPLL	21.31 / 0.6852	23.76 / 0.7317	25.31 / 0.7467	27.79 / 0.7759	29.83 / 0.8138	25.60 / 0.7507
	WNNM	22.19 / 0.6718	23.55 / 0.7293	25.23 / 0.7468	27.27 / 0.7560	29.67 / 0.7634	25.58 / 0.7335
	PURE-LET	22.42 / 0.6783	23.20 / 0.7344	25.95 / 0.7775	28.30 / 0.7511	29.54 / 0.7863	25.88 / 0.7455
Early DL	DnCNN	26.08 / 0.7499	27.89 / 0.7632	29.36 / 0.8021	30.67 / 0.7843	30.89 / 0.8444	28.98 / 0.7888
	IRCNN	25.86 / 0.7688	28.07 / 0.7883	28.60 / 0.7900	30.63 / 0.8349	30.74 / 0.8080	28.78 / 0.8000
	MemNet	25.23 / 0.7389	26.89 / 0.8134	28.33 / 0.7930	30.17 / 0.7802	31.77 / 0.8287	28.48 / 0.7948
SotA DL	Noise2Noise	26.80 / 0.7541	27.48 / 0.7748	29.16 / 0.7748	29.90 / 0.7799	32.35 / 0.8017	29.14 / 0.7771
	MWCNN	26.34 / 0.7448	27.48 / 0.7815	28.75 / 0.7654	30.61 / 0.8049	30.62 / 0.7794	28.76 / 0.7752
	RIDNet	25.86 / 0.7263	26.79 / 0.7908	28.90 / 0.7932	30.74 / 0.7909	30.73 / 0.8079	28.60 / 0.7818
	DPDN	26.87 / 0.7340	27.01 / 0.7608	28.96 / 0.8279	31.10 / 0.8155	32.11 / 0.8247	29.21 / 0.7926
	WF-UNet	25.40 / 0.6283	27.98 / 0.7034	29.26 / 0.7306	29.80 / 0.7487	30.22 / 0.7394	28.53 / 0.7101
	MSAN (Ours)	27.88 / 0.7820	30.24 / 0.8290	30.80 / 0.8338	32.76 / 0.8408	33.25 / 0.8330	30.98 / 0.8233

PSNR(dB)/SSIM
Methods		The number of raw images for averaging					Average value
Methods		1	2	4	8	16	Average value
Traditional method	Noised	20.67 / 0.2980	21.00 / 0.3917	23.00 / 0.4770	23.67 / 0.5559	23.75 / 0.6610	22.42 / 0.4767
	NLM	21.22 / 0.6428	23.03 / 0.7179	25.99 / 0.7328	27.88 / 0.7448	29.32 / 0.7985	25.49 / 0.7274
	BM3D	21.38 / 0.6941	24.47 / 0.7147	26.06 / 0.7499	28.43 / 0.7774	28.78 / 0.8093	25.82 / 0.7491
	KSVD	21.50 / 0.6635	24.02 / 0.6971	24.94 / 0.7588	26.90 / 0.7898	30.15 / 0.7898	25.50 / 0.7398
	EPLL	21.31 / 0.6852	23.76 / 0.7317	25.31 / 0.7467	27.79 / 0.7759	29.83 / 0.8138	25.60 / 0.7507
	WNNM	22.19 / 0.6718	23.55 / 0.7293	25.23 / 0.7468	27.27 / 0.7560	29.67 / 0.7634	25.58 / 0.7335
	PURE-LET	22.42 / 0.6783	23.20 / 0.7344	25.95 / 0.7775	28.30 / 0.7511	29.54 / 0.7863	25.88 / 0.7455
Early DL	DnCNN	26.08 / 0.7499	27.89 / 0.7632	29.36 / 0.8021	30.67 / 0.7843	30.89 / 0.8444	28.98 / 0.7888
	IRCNN	25.86 / 0.7688	28.07 / 0.7883	28.60 / 0.7900	30.63 / 0.8349	30.74 / 0.8080	28.78 / 0.8000
	MemNet	25.23 / 0.7389	26.89 / 0.8134	28.33 / 0.7930	30.17 / 0.7802	31.77 / 0.8287	28.48 / 0.7948
SotA DL	Noise2Noise	26.80 / 0.7541	27.48 / 0.7748	29.16 / 0.7748	29.90 / 0.7799	32.35 / 0.8017	29.14 / 0.7771
	MWCNN	26.34 / 0.7448	27.48 / 0.7815	28.75 / 0.7654	30.61 / 0.8049	30.62 / 0.7794	28.76 / 0.7752
	RIDNet	25.86 / 0.7263	26.79 / 0.7908	28.90 / 0.7932	30.74 / 0.7909	30.73 / 0.8079	28.60 / 0.7818
	DPDN	26.87 / 0.7340	27.01 / 0.7608	28.96 / 0.8279	31.10 / 0.8155	32.11 / 0.8247	29.21 / 0.7926
	WF-UNet	25.40 / 0.6283	27.98 / 0.7034	29.26 / 0.7306	29.80 / 0.7487	30.22 / 0.7394	28.53 / 0.7101
	MSAN (Ours)	27.88 / 0.7820	30.24 / 0.8290	30.80 / 0.8338	32.76 / 0.8408	33.25 / 0.8330	30.98 / 0.8233