Table 1.

Performance measures, (NRMSE, SSIM), for all reconstruction methods applied to raw data. The optimal number of iterations for L1 and LR were 110 and 150, respectively.

R=8
Subject	L1		LR		CD		CD+r (μ = 0.10)
	Performance	Regularization	Performance	Regularization	Performance	Number of steps	Performance	Number of steps
S1	(0.158, 0.862)	0.01	(0.141, 0.904)	0.0079	(0.139,0.894)	1702 (1500)	(0.149,0.883)	1414 (1500)
S2	(0.155, 0.864)	0.01	(0.137, 0.892)	0.0077	(0.134,0.876)	1695 (1500)	(0.136, 0.876)	1934 (2000)
S3	(0.135, 0.885)	0.01	(0.116, 0.921)	0.0073	(0.111, 0.914)	2060 (2000)	(0.111, 0.913)	2196 (2000)
R=12
Subject	L1		LR		CD		CD+r (μ = 0.10)
	Performance	Regularization	Performance	Regularization	Performance	Number of steps	Performance	Number of steps
S1	(0.179, 0.844)	0.01	(0.161, 0.882)	0.0063	(0.149,0.879)	1695 (1500)	(0.163,0.873)	1391 (1500)
S2	(0.177, 0.841)	0.0098	(0.155, 0.869)	0.006	(0.143,0.868)	1645 (1500)	(0.149,0.848)	1485 (1500)
S3	(0.157, 0.864)	0.0091	(0. 131, 0.904)	0.0058	(0.119, 0.907)	2156 (2000)	(0.120, 0.906)	1947 (2000)
R=18
Subject	L1		LR		CD		CD+r (μ = 0.10)
	Performance	Regularization	Performance	Regularization	Performance	Number of steps	Performance	Number of steps
S1	(0.210, 0.811)	0.0078	(0.189, 0.848)	0.0049	(0.162,0.880)	1798 (2000)	(0.198,0.840)	773 (1000)
S2	(0.207, 0.802)	0.0071	(0.182, 0.835)	0.0046	(0.151,0.848)	1728 (1500)	(0.163,0.850)	1440 (1500)
S3	(0.186, 0.836)	0.0065	(0.155, 0.877)	0.0045	(0.13,0.907)	2262 (2500)	(0.130, 0.907)	2967 (3000)
R=36
Subject	L1		LR		CD		CD+r (μ = 0.10)
	Performance	Regularization	Performance	Regularization	Performance	Number of steps	Performance	Number of steps
S1	(0.289, 0.726)	0.0038	(0.256, 0.775)	0.0048	(0.199, 0.837)	2177 (2000)	(0.339, 0.681)	9772 (9999)
S2	(0.277, 0.718)	0.0036	(0.249, 0.752)	0.0051	(0.185, 0.817)	2030 (2000)	(0.307, 0.701)	7660 (7500)
S3	(0.261, 0.756)	0.003	(0.219, 0.810)	0.0038	(0.155, 0.864)	2486 (2500)	(0.156, 0.866)	2551 (2500)