Table 2. Quantitative evaluation of the results of the different methods from Figure 8 for 50 K, 500 K, and 5 M counts.

Methods	50 K				500 K				5 M
	PSNR	SSIM	rRMSE		PSNR	SSIM	rRMSE		PSNR	SSIM	rRMSE
EM + Gaussian filter	27.81	0.686	0.515		31	0.896	0.349		37.726	0.967	0.232
EM + NLM	27.756	0.727	0.518		35	0.926	0.228		36.112	0.97	0.198
CNN	28.506	0.852	0.475		36	0.937	0.204		38.462	0.961	0.151
GAN	28.409	0.843	0.481		36	0.939	0.196		39.163	0.968	0.124
LCPR-Net	28.778	0.847	0.461		37	0.947	0.181		39.905	0.974	0.118

PSNR, peak signal-to-noise ratio; SSIM, structural similarity index; rRMSE, relative root mean square error; EM, expectation-maximization; NLM, nonlocal means; CNN, convolutional neural network; GAN, generative adversarial network; LCPR-Net, low-count positron emission tomography image reconstruction network.