Imaging quality of an artificial intelligence denoising algorithm: validation in 68Ga PSMA-11 PET for patients with biochemical recurrence of prostate cancer

EJNMMI Research

Table 5 Comparison of the signal-to-background ratios (SBR)

Series	Mean	Difference SBR	CI 95% of difference	p adjusted	Bias (%)
Signal/Background = SUVmax lesion/SUVmean aorta
QC4	5.28
SQC1	4.87	− 0.52	(− 1.22–0.3)	0.183	− 7.8
SQC2	4.37	− 0.94	(− 1.51–0.3)	0.009	− 17.2
SQC3	4.37	− 1.11	(− 1.58–0.62)	0.001	− 17.2
SQC4	4.28	− 1.16	(− 1.64–0.74)	< 0.001	− 18.9
VP4	5.19
SVP1	4.21	− 1.08	(− 2.22–0.07)	0.054	− 18.9
SVP2	4.11	− 1.29	(− 2.09–0.58)	0.007	− 20.8
SVP3	4.02	− 1.66	(− 2.13–0.56)	0.006	− 22.5
SVP4	4.17	− 1.27	(− 2.47–0.8)	0.002	− 19.6
Signal/Background = SUVmax lesion/SUVmean liver
QC4	1.13
SQC1	1.04	− 0.14	(− 0.29–0)	0.061	− 7.9
SQC2	0.92	− 0.19	(− 0.29–0.08)	0.002	− 18.5
SQC3	0.92	− 0.22	(− 0.36–0.14)	< 0.001	− 18.5
SQC4	0.88	− 0.29	(− 0.42–0.19)	< 0.001	− 22.1
VP4	1.07
SVP1	0.89	− 0.25	(− 0.52–0.05)	0.02	− 16.8
SVP2	0.85	− 0.22	(− 0.35–0.1)	0.007	− 20.6
SVP3	0.84	− 0.25	(− 0.47–0.16)	0.004	− 21.5
SVP4	0.85	− 0.28	(− 0.44–0.2)	0.001	− 20.6
Signal/Background = SUVmax lesion/SUVmean gluteal
QC4	14.89
SQC1	15.36	0.80	(2.06–2.94)	0.456	3.2
SQC2	13.60	− 1.01	(− 2.74–1.12)	0.247	− 8.7
SQC3	13.60	− 1.37	(− 2.34–0.2)	0.42	− 8.7
SQC4	13.29	− 2.07	(− 3.22–0.85)	0.13	− 10.7
VP4	13.96
SVP1	12.53	− 1.81	(− 3.58–0.92)	0.267	− 10.2
SVP2	12.91	− 0.88	(− 2.51–0.91)	0.214	− 7.5
SVP3	12.63	− 1.66	(− 2.65–0.36)	0.084	− 9.5
SVP4	12.67	− 1.76	(− 2.47–0.8)	0.17	− 9.2