PT - JOURNAL ARTICLE AU - Vy Le AU - Sarah Frye AU - Crystal Botkin AU - Kara Christopher AU - Praveen Gulaka AU - Barbara Sterkel AU - Ross Frye AU - Razi Muzaffar AU - Medhat Osman TI - <strong>Effect of PET Scan with Count Reduction Using AI-Based Processing Techniques on Image Quality</strong> DP - 2020 May 01 TA - Journal of Nuclear Medicine PG - 3095--3095 VI - 61 IP - supplement 1 4099 - http://jnm.snmjournals.org/content/61/supplement_1/3095.short 4100 - http://jnm.snmjournals.org/content/61/supplement_1/3095.full SO - J Nucl Med2020 May 01; 61 AB - 3095Objectives: Subtle Medical created the first Artificial Intelligent (AI) processing for F18-FDG (FDG) PET/CT imaging which was FDA approved in 2018. Subtle Medical specialized an AI processing technique for PET/CT imaging, called Subtle PET. The aim of Subtle Medical is to reduce noise resulting from time reduction and/or with radiopharmaceutical dose reduction. The purpose of this research is to evaluate and compare the image quality of study with standard full counts without AI processing and various count reductions with and without AI processing. Methods: The study was conducted with 15 FDG-PET/CT standard of care whole body studies from a VA population. The standard full count protocol is 3min/bed and the 50% count PET and 25% count PET were simulated to be 90 sec/bed and 45 sec/bed. Studies were reprocessed using different amount of counts: full count without AI processing, a 50% of total counts without AI processing, a 50% of total counts with AI processing, a 25% total of counts without AI processing, and 25% total of counts with AI processing. Scans were read and analyzed by two board certified nuclear medicine physicians. Image quality for each set of images was scored on a 5-points scale (1= poor and 5 = excellent image quality). BMI and age was also included in the analysis. Results: For the 15 patients, 5 image sets were evaluated and image quality was ranked (table 1). The mean image quality was as follows: 4.8 for full count scan without AI processing, 3.93 for the 50% which increased to 4.47 with AI processing, 3.00 for 25% total count which increased to 3.93 with AI processing. The data was also evaluated regarding impact of BMI on noise reduction using AI, with a BMI &lt; 30 (table 2) and BMI of 30 or greater (table 3). Conclusions: With 50% counts, Subtle PET AI processing improves image quality to that of the standard full counts, therefore, AI processing enables either 50% reduction in image acquisition or 50% reduction in FDG injected dose regardless of BMI. However, AI noise reduction is relatively less successful if producing high quality images if only 25% of the counts in used, particularly in patients with BMI &gt;30. Further studies are needed to test the success of AI in non FDG PET tracers and to elucidate the impact of AI on SUV quantification. Table 1: Diagnostic Image Quality of all PatientsView this table: Table 2: Diagnostic Image Quality BMI less than 30 View this table: Table 3: Diagnostic Image Quality BMI 30 and greater View this table: