%0 Journal Article %A Adam Riegel %A Osama Mawlawi %A Eric Rohren %A Homer Macapinlac %A Tinsu Pan %T Determination of PET thresholds for solitary pulmonary tumors in radiation treatment planning %D 2008 %J Journal of Nuclear Medicine %P 382P-382P %V 49 %N supplement 1 %X 1612 Objectives: To determine segmentation thresholds in PET images for solitary pulmonary tumors of varying sizes and different degrees of respiratory motion for radiation treatment planning. Methods: PET images of a regularly-oscillating NEMA IEC phantom were acquired using a GE Discovery VCT scanner. Scans of the 6 spheres (diameters of 1.0 to 3.7 cm) in the IEC phantom were taken at different extents of motion ranging from 0.5 cm to 3 cm at 0.5 cm intervals. A range of image thresholds (5% to 99%) were applied to the PET images to determine which threshold produced volumes closest to the ideal volume (that of a sphere plus a cylinder of height equal to the motion extent). PET thresholds were based on maximum SUV. A look-up table with thresholds for solitary pulmonary nodules of different degrees of motion and various sizes was established for determining tumor volume in clinical PET images. Lesion size and motion extent were determined for the solitary pulmonary nodules of 6 patients in 4D-CT. Volumes resulting from PET thresholding with the look-up table were compared with volumes from 4D-CT. Results: The thresholds varied considerably with object size and motion extent. When applied to patients, the average percent difference in volume between 4D-CT and PET contours for the 6 patients was less than 10%. Conclusions: As previous research suggests, one threshold is not adequate for segmenting objects of varying degrees of size or motion. The threshold look-up table presented here has produced PET volumes that are similar to volumes produced with 4D-CT image sets. Research Support: Schissler Foundation M. D. Anderson Cancer Center Fellowship %U