Benefit of time-of-flight imaging for low BMI patients in F-18 FDG PET/CT

1473 Objectives: Time-of-flight (TOF) information in PET data has been shown to provide improved signal-to-noise ratio in the reconstructed images while also minimizing artifacts due to inconsistencies in the data. In the case of large patients, the use of TOF reconstruction has been shown to provide visually better image quality compared to non-TOF reconstruction. However, there has been very little evidence of the benefit of TOF in low body mass index (BMI) patients. The aim of this work was to investigate the benefit of using TOF reconstruction for patients with BMI less than 20 and compare with the corresponding non-TOF reconstruction for routine F-18 FDG PET/CT scans. Methods: 50 patients undergoing F-18 FDG PET/CT scan as part of the diagnostic protocol and having a BMI of less than 20 were recruited for this study. A whole-body PET/CT scan was performed for each patient (45 minutes p.i) on the Siemens Biograph mCT 3-ring system using the step-and-shoot acquisition with a scan duration of 40 seconds per bed. Each dataset was processed using non-TOF reconstruction (2 iterations, 24 subsets and 3.5mm Gaussian filter) for voxel sizes of 4.07x4.07x2.027mm. In addition, TOF reconstruction (2 iterations, 21 subsets and 3.5mm Gaussian filter) was performed with the same voxel sizes where the reconstruction parameters for TOF processing were chosen in order to match the image noise observed with non-TOF images. Three nuclear medicine physicians, blinded to the type of reconstruction, identified the lesions on all the whole-body PETCT images. Quantitative measures of the SUVmax and SUVmean were computed for all the lesions identified. PET image based tumor volumes were computed for all lesions as the isocontour volume using 40% of SUVmax as the threshold. In addition, qualitative metrics features for each image such as lesion conspicuity, liver uniformity, organ detail and overall diagnostic confidence based on impression of contour/smoothness/homogeneity/image artefact were scored by the physicians on a scale of 0 - 3 where 0 was worst and 3 was the best. The noise measures for both sets of reconstruction was quantified by computing the normalized standard deviation in a matched liver region of interest. Results: A total of 87 lesions were identified on the PET/CT images for all the patients. The noise measure in the liver volume of interest was found to be similar between nonTOF (13.5%) and TOF (13.9%) reconstruction. On average the SUV measures with TOF reconstruction were higher (SUVmax was 14.6% higher and SUVmean was 15.3% higher) than those obtained with nonTOF reconstruction. Large differences in the estimated PET tumor volumes were found between TOF and nonTOF reconstruction with an average absolute difference in volume of 31.3% ± 25. In particular, the differences were found to be greater for SUVmean measures less than 10. Average qualitative measures were all found to be marginally higher for TOF reconstruction compared to nonTOF reconstruction (lesion conspicuity = 2.62 vs 2.47, liver uniformity = 2.63 vs 2.49, organ detail = 2.65 vs 2.55 and overall diagnostic confidence based on impression of contour/smoothness/homogeneity/image artefact = 2.68 vs 2.59). Conclusions: Our work of evaluating TOF reconstruction for low BMI patients demonstrated higher SUV measures with TOF reconstruction compared to nonTOF reconstruction which were likely more accurate due to better convergence with TOF reconstruction. In addition, higher image quality metrics were obtained for TOF reconstruction compared to nonTOF reconstruction. These findings suggest that the use of TOF reconstruction in subjects with low BMI provides improved image quality and may potentially help in better patient management.