Quality assessment of quantitative values on dedicated-breast PET

1467 Objectives To evaluate the quality of quantitative values on our dedicated-breast PET scanner (DB-PET) comparing to whole-body PET (WB-PET). Methods First, a pool phantom filled with FDG solution was scanned by a DB-PET scanner. Standardized uptake values (SUV) were obtained at 5-mm intervals from end to end of the gantry of 15-cm depth using ROI of 50 cm2. Standard deviations (SD) of SUVs were evaluated according to the depth. Second, a breast-simulate phantom with 10-, 13-, 17-mm hot spheres (sphere-to-background activity ratio = 4.3) were positioned at 75-mm (center) or 37-mm (off-center) from the the end, and scanned with gating using WB-PET/CT and DB-PET scanners. All images were interleaved, and two datasets of images were obtained for each scanning. The percent difference (d) of SUVmax between the two datasets was calculated as following formula; d = (SUVmax2 - SUVmax1)/SUVmax1 × 100, where SUVmax1(2) was determined on a PET image reconstructed by odd(even) image data. All images were reconstructed with attenuation correction using 3-min and 5-min acquisition data for WB-PET and DB-PET, respectively. Results In the pool phantom study, mean SD was 0.029±0.005. Highest SDs were observed at both end; 0.049 and 0.041 at 5-mm level from the ends. In the breast-simulate phantom study, mean absolute difference was 7.1±4.9% in WB-PET and 3.2±5.4% in DB-PET (P = 0.03). The DB-PET tended to provide relatively low difference in large and medium spheres. Comparing to large spheres, SUVmax of medium and small spheres decreased to 94.6±6.8% and 75.2±7.6% in WB-PET and to 98.1±3.5% and 82.3±4.8% in DB-PET, with the statistical significance in small spheres (P = 0.01). Conclusions Although the depth-dependency may exist, our DB-PET is potentially useful to provide SUVmax, which has better reproducibility and is less susceptible to the partial volume effect in small tumors, comparing to WB-PET.