Abstract
P613
Introduction: Long axial field of view (LAFOV) PET/CT has enabled increased image quality and lesion detection, while maintaining shorter total scan time than previous standard field of view PET scanners. Further, reduction of injection dosage is made possible, while still upholding image quality of current clinical practice, thus minimizing radiation exposure. In the summer of 2022, the clinical pediatric imaging LAFOV PET/CT protocol was updated at our facility, lowering administered FDG-dosage. In this study, we aim to evaluate the imaging quality and diagnostic performance during the clinical implementation of FDG-dosage reduction from 3 to 1.5MBq/kg. We investigate the clinical consequences of various settings concerning filter application, number of iterations and acquisition time, through measurements of image noise as well as reports of lesion detection and classification.
Methods: This is a retrospective quality control study, performed on a LAFOV Siemens Biograph Vision Quadra PET/CT scanner. The first 30 pediatric patients referred for whole-body FDG-PET/CT after implementation of LAFOV PET/CT with lowered FDG dose (1.5MBq/kg), were included. Imaging protocol included; 10min acquisition time, 4-6 iterations (i), 5 subsets, possible Gauss filter of 2mm, PSF and TOF modeling. Five reconstructions were then recreated from the list mode data stream from scan start until the given time point and included in the systematic evaluation; (A) Gauss Filter, 4i, 150s (F150); (B) Gauss Filter, 4i, 300s (F300); (C) Gauss Filter, 4i, 600s (F600); (D) No Filter, 4i, 300s (N4i300) and (E) No Filter, 6i, 300s (N6i300). Image quality computed as coefficient of variance (COV) was compared from 10-25mm radius (age dependent) VOI’s, placed in the right liver lobe. Two nuclear medicine physicians with +5 years of PET experience were randomly assigned sets of scans consisting of the PET reconstructions (A-E) along with the CT. All image lesions and classification were reported on a predefined list of 43 anatomical sites. Analysis of variance and paired samples comparisons, Bonferroni corrected by a factor of 12 (PB), were performed.
Results: All acquisitions achieved a COVmean of 0.09-0.18. The following reconstructions; F300, F600 and N4i300 were the only ones to achieve a COVmean within current acceptable clinical values at our institution (COVref ≤ 0.15), with the F600 providing least image noise and the N4i300 the most. COV was found to be significantly different across acquisitions (F(2.07,60.00)=713.53, p<0.001), with significant improvement from N6i300 to N4i300 to F300 as well as with increasing acquisition time from F150 to F300 to F600 (PB<0.001). Significant difference was also seen in lesion detection across acquisitions (F(1.11,46.74)=7.86, p=0.006). A significant increase was seen from F150 to F300 and from F150 to F600 (PB=0.012), but not from F300 to F600 (PB=0.24). Likewise no significant difference in lesion detection was observed between the three acquisitions (F300, N4i300, N6i300) comparing filter and iteration settings. Further no change in lesion classification was observed at any point.
Conclusions: Image quality improved significantly with lowering the number of iterations from 6 to 4, by applying the Gauss filter and with increasing acquisition time. While acceptable values for image noise could be achieved after 150s, a significant improvement in diagnostic performance, measured as lesion detection, was observed by increasing acquisition time to 300s or 600s. No significant difference in lesion detection was observed from 300s to 600s, when removing the Gauss filter or changing the number of iterations at 300s. Further, no difference in lesion classification was observed. These results find a clinical pediatric imaging LAFOV PET/CT protocol of injected FDG dose of 1.5MBq/kg, 2mm Gauss filter, 4i and 300s acquisition time provides a diagnostic performance of high clinical quality.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.