TY - JOUR T1 - Comparison between SiPM-based and conventional PMT-based PET scanners in different reconstruction algorithm using ACR phantom JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 3089 LP - 3089 VL - 61 IS - supplement 1 AU - Min-Tzu Ku AU - Yi-Lun Chen AU - Bang-Hung Yang AU - Wen-Sheng Huang AU - Ren-Shyan Liu Y1 - 2020/05/01 UR - http://jnm.snmjournals.org/content/61/supplement_1/3089.abstract N2 - 3089Introduction: Digital PET scanner using silicon photomultiplier is considered to have higher detection capability than analog PET. In addition, a novel PET reconstruction algorithm BSREM (Q.Clear®, GE Healthcare, Waukesha, WI) controls signal at higher levels while mitigating noise amplification, correcting the problem that standard OSEM reconstruction algorithm always have. In this study, we compared maximum SUV (SUVmax) and image quality between digital and analog PET scanners with standard OSEM and BSREM reconstruction. Materials and Methods: The ACR phantom was scanned on GE Discovery MI (digital PET/CT with TOF), Discovery MIDR (analog PET/CT with TOF) and SIGNA (digital PET/MRI with TOF) using a standard 4 minutes long acquisition protocol. The faceplate has four fillable hot cylinders (8, 12, 16 and 25 mm in diameter) and three cold cylinders (air, water and bone equivalent material). According to ACR Nuclear Medicine Accreditation Program-PET Module, we generated the 2.5 concentration ratio and selected patient dose 10 mCi protocol to fill the phantom. The dose of 18F-FDG was diluted to the same concentration as the protocol, and the phantom scan was obtained 1 hour after dose was measured. The standard OSEM algorithm available on the scanners with 3 iterations and 16 subsets for Discovery MI, 2 iterations and 24 subsets for Discovery MIDR, and 2 iterations and 28 subsets for SIGNA (as recommended by the manufacturer). All of them used 5.0 mm filter. The BSREM algorithm with 25 iterations and different regularization strength parameters (β value between 200 and 850) without any filter applied. AW sever was used for data analysis to draw ROI and calculate SUV values. Results: SUVmax measuerd by digital and analog PET scanners are different. The SUVmax in Discovery MI and SIGNA are obviously higher than Discovery MIDR and is independent of lesion size. Second, images reconstructed by BSREM show better image quality with low noise than standard OSEM in three PET scanners. Furthermore, SUVmax also increase while using BSREM reconstruction, and is more significant in small lesion. Last, data illustrate that the BSREM algorithm regularization strength parameters (β value) in Discovery MIDR should be selected under 500, or SUVmax would be lower than using standard OSEM algorithm. Otherwise, while using Discovery MI and SIGNA, higher β value should be chose to enhance both image quality and SUVmax. Conclusions: Our preliminary data showed that BSREM algorithm is advantageous on PET images, especially in small lesion, but the optimal regularization strength parameters (β value) must be defined. The selection of β value depend on lesion size and different type of PET scanner. It is necessary to modified suitable conditions in both digital and analog PET scanners to improve image for clinical use. ER -