Direct comparison of [ 18 F]FDG brain images acquired from a phantom and patients using SiPM- and PMT-PET/CT CURRENT STATUS: UNDER REVIEW

image quality, uptake value ratio, Abstract Background Silicon photomultiplier-positron emission tomography (SiPM-PET) has better sensitivity, spatial resolution, and timing resolution than photomultiplier tubes (PMT)-PET. The present study aimed to clarify the advantages of SiPM-PET in 18 F-fluoro-2-deoxy-D-glucose ([ 18 F]FDG) brain imaging in a head-to-head comparison with PMT-PET in phantom and clinical studies. Methods Image contrast was calculated from images acquired from a Hoffman 3D brain phantom and image noise and uniformity were calculated from pooled images acquired from a pool phantom using SiPM-and PMT-PET. Sequential PMT-PET and SiPM-PET [ 18 F]FDG images were acquired over a period of 10 min from 22 individuals. All images were separately normalized to a standard [ 18 F]FDG PET template, then mean standardized uptake values (SUV mean ) and Z-score were calculated by MIMneuro and Cortex ID Suite, respectively. The SUV mean of all regions for SiPM- and PMT- PET acquisitions were statistically compared using two-tailed paired Student t tests. Spearman rank correlation coefficients were calculated to evaluate relationships among different SUV mean in the whole brain and intervals between acquisitions. Z-scores were statistically analyzed for both acquisitions using Wilcoxon matched-pairs signed rank tests. Values with P < 0.05 were considered significant. Three-dimensional stereotactic surface projections

based on the statistical image analysis because the SiPM-PET was more localized the distribution of glucose metabolism on Z-score maps.

Introduction
Positron emission tomography (PET) has become an important imaging technology for evaluating biochemical and physiological functions and pathological abnormalities (1), (2). Brain imaging with 18  SiPM-and PMT-PET scanners (7). Image contrast, noise, and spatial resolution were better for images acquired using digital PET in their clinical study.
The present study aimed to clarify the advantage of SiPM-PET system in [ 18 F]FDG brain imaging in head-to-head comparisons between DMI and D710 in phantom and clinical studies. To our knowledge, this is the first attempt to evaluate the image quality with phantom study and quantitative value and the results of statistical image analysis with clinical study in SiPM-PET.

Discovery MI
The Discovery MI is a combination of LBS, an SiPM-PET scanner and a 64-slice CT scanner. The LBS includes four blocks of detectors aligned in the axial direction, each comprising 19,584 crystals subsets; Gaussian filter, 2.5 mm (FWHM); 128 × 128 matrix size; FOV, 25.6 cm; 2.0 mm/pixel. The PET images of the Hoffman 3D brain phantom acquired by both scanners were reconstructed as described without TOF to evaluate TOF gain in background counts.

Data acquisition
Images from Hoffman 3D brain phantom (Data Spectrum Corporation, Hillsborough, NC, USA) and pool phantom (Itoi Plastics Co. Ltd., Kobe, Japan) containing 20 MBq of [ 18 F]FDG were acquired for 30 min in list mode using the SiPM-PET and PMT-PET systems. Phantom conditions and the scan duration were determined according to the Japanese Society of Nuclear Medicine (JSNM) phantom test procedure (9). We extracted a time frame of 0 -7 min from 30 min of data derived from the Hoffman and pool phantoms that was equivalent to the count statistics for [ 18 F]FDG clinical brain images at the Tokyo Metropolitan Institute of Gerontology (TMIG) as described below.

Data processing
The physical indices for phantom tests proposed by the JSNM were used to evaluate the image quality: the ratio of grey-to-white matter contrast (contrast) calculated from images of Hoffman phantom, image noise (coefficient of variation, CV [%]) and uniformity (standard deviation, SD) calculated from images of pool phantom (9). The SD was also calculated from the pool phantom image with a scan duration of 30 min. The contrast, CV and SD were respectively calculated as described using images acquired from Hoffman and pool phantoms (9). Eight 10-mm circular volumes of interest (VOI) were placed on images of the acrylic plate at the bottom of the Hoffman phantom that were reconstructed using 3D-OS-EM with and without the TOF in the background (BG) (Fig. 1).
The TOF gain (%) in the background counts was calculated as:

Data acquisition
The present study proceeded in accordance with the Declaration of Helsinki, and was approved by the Ethics Committee at the TMIG (Approval No. 28077). All applicants provided written informed consent to participate in the present study after physicians provided a detailed explanation of the study. The individuals rested comfortably in a quiet, dimly-lit room for several minutes, then were

Results
8 Phantom study Table 2 shows that the physical indices of SiPM-PET satisfied the JSNM image quality acceptance criteria of contrast > 55%, CV ≤ 15% and SD ≤ 0.0249 (the clinical protocol at TMIG), whereas the contrast of PMT-PET images acquired for 7 min did not. The uniformity of the pool phantom images acquired for 7 min using SiPM-PET was also adequate. The BG count was dramatically reduced by using the SiPM-PET with TOF.  Clinical study Figure 2 and Table 3 show changes in SUV mean and mean (± SD) SUV mean across all brain regions, respectively, between acquired using SiPM-and PMT-PET. The SUV mean was significantly higher on SiPM-PET than PMT-PET images in all region. The mean (± SD) of the interval between sequential acquisitions (PMT-PET followed by SiPM-PET) was 15.2 ± 1.0 min. The second acquisition started about 5 min after the end of the first acquisition. Figure 3 shows correlations between changes of SUV mean in whole brain and time between first and second acquisitions. The R of the SUV mean was 0.06 (P = 0.79), then the SUV mean was independent of the time. Data are shown as means ± standard deviation. PET, positron emission tomography; PMT, photomultiplier tube; SiPM, Silicon photomultiplier. Figure 4 shows that the comparisons of Z-scores that was analyzed using CotexID Suite in all regions (except the bilateral posterior cingulate) were significantly higher in SiPM-PET than PMT-PET images. where were corresponded to the cerebellum (− 40 mm) and parietal lobe (+ 40 mm) in the human brain (9). Uniformity can be estimated as an index of the count stability through the entire axial FOV.
Good uniformity means the less statistical noise on PET image at the edge of axial FOV. The SiPM-PET could include a whole brain within its PET axial FOV. The statistical noise was suppressed at the bottom of the brain such as pons and cerebellum where were the reference region to calculate the SUVR for [ 18 F]FDG (10), amyloid (14,15), and Tau (16) PET imaging using the SiPM-PET. Therefore, the SUVR calculated by SiPM-PET was expected to be stable.
The SiPM-PET images acquired from Hoffman phantom had good image contrast and decreased residual BG counts due to good spatial and timing resolution. The improvement in image contrast (38.0%) were the same in the phantom study as that clinical study by Philips SiPM-PET (7). However, Salvadori et al. did not find a benefit of TOF with a digital PET system for brain PET (7). The timing resolution of the DMI and D710 was 375 and 544 ps, respectively (4,8). These led to spatial localization along a line of response of 5.8 and 7.5 cm, respectively (3). The sensitivity gain using TOF was increased as a function of increasing the object size (17). Nagaki et al. found that contrast in [ 18 F]FDG brain imaging is not improved using the PMT-PET system at a timing resolution of 555 ps (18). The SiPM-PET with TOF improved image contrast even for small objects such as a human brain compared with a human body.
The clinical study showed that the SUV mean was significantly higher using SiPM-PET than PMT-PET and did not correlate with the delay of the time from the injection (19). The superior spatial and timing resolution by SiPM-PET not only improved image contrast but also increased the SUV mean in the cortex (2). The higher Z-scores determined using SiPM-PET was affected by higher SUV mean and lower image noise. Cortex ID Suite uses the three-dimensional stereotactic surface projections (3D-SSP) as a method of statistical image analysis (11). The SiPM-PET raised the peak signal on the cortex that was used to analyze the 3D-SSP because small scintillator crystals in the SiPM-PET reduced the partial volume effects in the signal of grey matter (4). Salvadori et al. also found better recovery coefficient Figure 1 Position of circular ROI to measure background counts on bottom of Hoffman 3D brain phantom. Four circular ROI were placed on acrylic plate background and others were placed next to slices. ROI, regions of interest.