Application of Compressive Sensing to Ultrasound Images: A Review

Table 2

CS reconstruction algorithm based on sparsifying transforms.

S/N	Reference	Method	Average MAE
1	[39]	Model basis; (i) Wavelet atom; (ii) D cosine T; (iii) DFT reconstruction; (iv) L1 minimization; (v) BSB	L1-FT		1.8994e − 04
			L1-DCT		1.3124e − 04
			L1-WA		1.2161e − 04
			BSBL-FT		1.3693e − 04
			BSBL-DCT		9.5381e − 05
			BSBL-WA		1.6805e − 04
2	[37]	Model basis; (i) Directional wave atoms; (ii) Daubechies wavelets; (iii) Fourier transform	L1-wavelet		1.5163e − 03
			L1-DCT		8.3572e − 04
			L1-W atom		5.5428e − 04
3	[38]	Model basis; (i) DWT; (ii) DCT Reconstruction; Convex optimization; L1 minimization	MSE
			CS-flow 1			2.34
			CS-flow 2			2.95
			CS-flow 3			4.34
4	[40]	Model basis; (i) curvelets; (ii) Wavelet; (iii) Cosine; (iv) Fourier	Method	PSNR				SSIM
			Cyst phantom image
			Frequency domain	25.758			0.726
			Time domain	22.857			0.701
			Liver image
			Frequency domain	32			0.783
			Time domain	20.2			0.741
5	[32]	(i) Approximate messaging passing model basis; (ii) DCT; (iii) Wavelet; (iv) Spatial domain ST and ABE as denoiser	Time ST	9.09			0.14
			Time ABE	8.57			0.09
			Wavelet ST	12.46			0.28
			Wavelet ABE	12.38			0.25
			DCT ST	18.56			0.54
			DCT ABE	23.95			0.80