Effect of different respiratory motion correction methods on PET image quality in chest PET/MRI

Abstract

Objective: To investigate the effect of different respiratory motion correction methods on PET images during chest PET/MRI scans. Methods: The data of 35 patients (24 males and 11 females, aged from 29 to 84 year) of pulmonary lesions with significantly high uptake in thoracic PET/MRI scan were retrospective collected from Jingling Hospital. Four different methods were used to reconstruct the PET data. Group A was the full-time 20 min without respiratory motion correction static acquisition (Static) as a control, group B was the end-expiration static collection (Q.Static), and group C was the multi-bins respiratory gating (Gated-Respiratory). In addition, the influence of the time being considered, group D was added for reconstruction in the first 1/3 period (6 min 40 s) of group A. Then, the maximum value (L-SUVmax) and the mean value (L-SUVmean) of the SUV of the lesion, the mean value (B-SUVmean) and the standard deviation (B-SUVsd) of the SUV of the background under each reconstruction results were measured, and for each lesion the signal-to-noise ratio (L-SNR) was calculated. In order to exclude the interference of the background, the mean of the relative SUV (L-dSUVmean) of the L-SUVmean relative to the B-SUVmean was also calculated. Finally, One-Way Repeated Measures ANOVA was used, and the post-hoc pairwise comparison between groups was tested by Bonferroni's modified test. Results: There was statistically significant difference among group B or group C compared to group A and group D in L-SUVmax, L-SUVmean and L-dSUVmean [L-SUVmax:group B vs group A or group D was 8.06±3.57 vs 7.73±3.45 or 7.61±3.50, group C vs group A or group D was 8.04±3.56 vs 7.73±3.45 or 7.61±3.50 (all P<0.05); L-SUVmean: group B vs group A or group D was 4.12±1.78 vs 3.98±1.72 or 3.91±1.71, group C vs group A or group D was 4.13±1.78 vs 3.98±1.72 or 3.91±1.71 (all P<0.05); L-dSUVmean: group B vs group A or group D was 3.52±0.16 vs 3.39±0.18 or 3.31±0.18, group C vs group A or group D was 3.53±0.18 vs 3.39±0.18 or 3.31±0.18 (all P<0.05)], but there was no statistically significant difference between group B and group C (all P>0.05). There were statistically significant differences between group D and group A in B-SUVsd (0.07±0.00 vs 0.07±0.00, P=0.023) and L-SNR (69.80±44.57 vs 85.35±68.98, P=0.001). There was no statistically significant difference between group D and group A in L-SUVmax, L-SUVmean, B-SUVmean and L-dSUVmean (all P>0.05). Conclusions: There was no significant difference in PET image quality between the Q.static and Gated-Respiratory group, both of the two groups were better than the Static group which with no Gated-Respiratory motion correction. If non-respiratory gated Static is used, the PET acquisition time is recommended to be 6 min 40 s.