TY - JOUR T1 - Variations in SUV values with reconstruction algorithms JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 2100 LP - 2100 VL - 52 IS - supplement 1 AU - Bing Bai AU - Michal Kulon AU - Peter Esser Y1 - 2011/05/01 UR - http://jnm.snmjournals.org/content/52/supplement_1/2100.abstract N2 - 2100 Objectives We compared the mean and maximum standard uptake values (SUVs) measured from PET images reconstructed with FBP, and OSEM, with and without point spread function (PSF) modeling, using both phantom and patient data. Methods A 14 cm inner diameter (ID) cylindrical phantom with 7 fillable spherical inserts (ID 0.5-3.1 cm) was scanned using a 64 slice PET/CT scanner. A circular 2D region of interest (ROI) was drawn manually at the center of each sphere in the transaxial slice on the coregistered CT image. Activity was measured using a dose calibrator to calculate true SUV values (2.57/1.03 sphere/background). Three patient PET/CT scans with well-delineated lung tumors were reconstructed. ROIs were drawn based on a clinical protocol which selects the voxels exceed 50% of the maximum SUV in each tumor. Results Differences between FBP and OSEM were small for all measurements. PSF-OSEM had higher SUV values related to the resolution recovery and the edge artifacts, that enhance the image values near the sharp edges in the image. The maximum SUV from phantom PSF-OSEM image may over-estimate the value (up to 9%, measured with the 1.5 cm ID sphere). For patient data, the PSF-OSEM SUVs were considerably higher (10-20%) than FBP and OSEM. The relative differences of mean SUVs between PSF-OSEM and FBP were larger for patient data than the phantom data, probably as a result of the different ROI sizes with each reconstruction algorithm. Conclusions Mean and maximum SUV measured from FBP and OSEM images were similar, while PSF-OSEM values were significantly higher. Based on this data, for tumors between 1-3 cm in diameter, using the maximum SUV may overestimate the SUV value with PSF-OSEMMean and Maximum SUV of patient data ER -