Analysis of PET images obtained by a newly-developed mobile PET scanner according to three different layouts of detectors: reconstruction using NLM method.

572 Objectives: A mobile PET scanner (fxPET) has been developed to fit the existing MRI or CT system. The non-local means (NLM) is a novel smoothing and feature-preserving method for image reconstruction. The purpose of this study was to evaluate NLM-based fxPET images in terms of the detection rate of pathological lesions, image quality, and the quantitative values of three different layouts of detectors, comparing with those of conventional whole-body (WB) PET/CT. Methods: 96 patients with various malignancies underwent WB PET/CT scan first after injection of 18F-FDG (n=86) or 68Ga-DOTATOC (n=10), followed by fxPET scanning with detectors oppositely located above and below patients (Layout A). Of 96 patients, 70 patients had additional fxPET scanning with detectors more closely (Layout B) in Group 1. The remaining 26 patients had additional fxPET scanning with detectors set bilaterally (Layout C) in Group 2. Attenuation correction for fxPET images was performed using CT obtained by WB PET/CT. The lesion-based detection rate in each layout was compared with that of WB PET/CT. For image quality, two readers rated each image on consensus by using 4-point grading scale (3: perfectly demarcated without distortion, 2: partially mixed with background or faintly distorted, 1: moderately distorted or ill-defined, 0: inappropriate to diagnosis due to degraded image). We calculated the SUVmax/mean/peak, MTV, and TLG in three layouts of fxPET using FDG in Group1 (n=63) and Group 2 (n=23), and compared them with those in WB PET/CT. Additionally, we compared the signal-to-background ratio (SBR) between each layout and WB PET/CT. In the neck-pelvic, the thoracic, and the upper abdominal regions, the SBR was defined as SUVmax of the lesion divided by the SUVmean of the muscle, the descending aorta, and the normal liver, respectively. Results: The detection rates are demonstrated in Tables 1 and 2. In Group 1, the detection rates with Layout A in nodal metastases and those with Layout B in nodal and liver metastases were better than those of WB PET/CT (p<0.001). In Group 2, the detection rates with Layout A and C was comparable with that in WB PET/CT. As for the image quality, there were 16 (17%), 78 (81%), 2 (2%), and 0 (0%) cases, in grade 3, 2, 1, 0 respectively, in Layout A, 45 (64%), 25 (36%), 0 (0%), and 0 (0%) cases, respectively, in Layout B, and 2 (8%), 23 (88%), 1 (4%), and 0 (0%) cases, respectively, in Layout C. In Group 1, the number of cases of grade 3 was significantly larger in Layout B than in Layout A (p<0.0001), while there was no significant difference between Layout A and Layout C in Group 2. All the quantitative values obtained with three layouts using FDG had strong positive correlation with those in WB PET/CT (rho: 0.80 - 0.92, p<0.0001). All the SBR of three layouts was larger than in WB PET/CT, and was statistically significant (p<0.05) other than that of neck-pelvic and upper abdominal regions of Layout A and C in Group 2. Conclusions: In fxPET scanning, higher detection rate and image quality were obtained in Layout B. All the quantitative values in three layouts of fxPET had significantly strong positive correlation with those in WB PET/CT, suggesting clinical usefulness and feasibility of fxPET. All the SBR of three layouts was larger than in WB PET/CT, possibly due to the delayed scan time. This presentation is based on results obtained from a project commissioned by the New Energy and Industrial Technology Development Organization (NEDO).