Learned Shrinkage Approach for Low-Dose Reconstruction in Computed Tomography

Table 1

Quantitative measures for the compared algorithms.

Method	FBP	ATM	PWLS	Shrinkage
Thighs
MSEg =
MSE + grad	16.12 + 11.64	11.81 + 12.54	7.98 + 5.31	6.99 + 11.17
MSEg total	27.76	14.35	13.29	18.16
SNR (dB)	26.56	28.70	29.56	30.73
SSIM	0.8643	0.9124	0.9219	0.9398
Abdomen
MSEg =
MSE + grad	14.96 + 6.88	11.78 + 6.73	7.12 + 7.49	6.80 + 7.18
MSEg total	21.83	18.52	14.61	13.99
SNR (dB)	26.42	27.34	29.53	30.41
SSIM	0.8832	0.8985	0.9326	0.9483
Head
MSEg =
MSE + grad	6.99 + 11.40	6.99 + 11.39	3.31 + 7.99	2.77 + 6.38
MSEg total	18.40	18.38	11.30	9.15
SNR (dB)	30.84	30.84	31.94	33.74
SSIM	0.9764	0.9764	0.9783	0.9864