PT - JOURNAL ARTICLE AU - Jason Parker AU - Senait Debebe AU - Wendy Territo AU - Harshali Bal AU - Maurizio Conti AU - James Fletcher AU - Gary Hutchins TI - Clinical evaluation of Poisson maximum likelihood-based scatter subtraction for <sup>68</sup>Ga-PSMA imaging DP - 2020 May 01 TA - Journal of Nuclear Medicine PG - 107--107 VI - 61 IP - supplement 1 4099 - http://jnm.snmjournals.org/content/61/supplement_1/107.short 4100 - http://jnm.snmjournals.org/content/61/supplement_1/107.full SO - J Nucl Med2020 May 01; 61 AB - 107Introduction: Robust scatter correction is necessary for accurate clinical PET quantification, particularly for challenging tracers such as gallium-68-labeled prostate-specific membrane antigen (68Ga-PSMA) which are characterized by high uptake throughout the urinary system. Halo artifact associated with overcorrection of scatter fields in the vicinity of the urinary system organs - notably the kidneys, ureters, and bladder - has been observed and is recognized as an obstacle for the clinical acceptance of 68Ga-PSMA PET [1, 2]. Recent techniques for scatter subtraction based on a Poisson maximum likelihood-based solution with nested scatter scaling (MLSS) estimation and activity image update have shown improved qualitative and quantitative image quality in phantom and limited patient studies [3-5]. However, these methods have not been thoroughly tested in a realistic clinical patient population. The purpose of this work was to evaluate the image quality, quantification, and lesion detection characteristics of an MLSS technique used to reduce halo artifact in a clinical 68Ga-PSMA PET population. Methods: Seventy-six (76) prostate cancer patients underwent whole body 68Ga-PSMA scanning on a Biograph mCT PET/CT scanner (Siemens Healthineers, Siemens USA). Raw list mode data were reconstructed using OSEM (21 subsets, 2 iterations) with 3 different scatter correction methods: 1) absolute single scatter scaling (SSS), 2) relative SSS, and 3) MLSS. CT-based µ-maps were calculated and used for attenuation correction and anatomic localization. Scatter fractions (SF) and lesion standardized uptake values (SUV) were calculated for all patients and compared across the 3 correction methods. Scatter profile analyses were conducted by summing 10 slices across the bladder of each patient. SUVs in normal (adjacent to bladder), lesion, and bladder ROI’s were compared across the 3 methods. Paired t-tests were used to evaluate statistical differences in all measures. Results: MLSS demonstrated increased estimated SF compared to absolute SSS (p&lt;10-8) and decreased SF compared to relative SSS (p&lt;10-8). These results agree with previous phantom studies. MLSS demonstrated increased SUVs in normal ROI’s compared to relative SSS (p&lt;10-8) and decreased SUVs in normal ROI’s compared to absolute SSS (p&lt;10-8). These results are consistent with reduced halo artifact in the vicinity of the urinary bladder. There were no significant differences in SUV in the lesion or bladder ROI’s. Conclusions: MLSS reduces halo image artifacts in regions adjacent to areas with high radionuclide accumulation levels and demonstrated statistically significant differences in estimated scatter fractions between methods consistent with previous phantom studies. Qualitatively, MLSS appears to favorably alter the appearance of 68Ga PSMA PET images of the pelvis and may result in improved detection of abnormal peri-cystic foci of tracer uptake. An ongoing study is evaluating lesion conspicuity between the three methods.