TY - JOUR T1 - Validation of quantitative tumor imaging with small-animal PET JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 396P LP - 396P VL - 49 IS - supplement 1 AU - James Bading AU - Jiping Zhan AU - Lawrence Williams Y1 - 2008/05/01 UR - http://jnm.snmjournals.org/content/49/supplement_1/396P.2.abstract N2 - 1668 Objectives: Few studies have validated the accuracy and precision of small-animal PET for measurement of absolute radioactivity concentrations. We compared image-derived and ex vivo assay measurements of tumor activity concentrations in mice. Methods: Mice bearing superficial tumors (17-227 mg, n=21), were injected with [18F]FDG or various 64Cu-labeled radiotracers and imaged with an R4 microPETTM scanner. Mice were then immediately killed, and tumors were excised and assayed for activity concentration (AC). MicroPET raw data were corrected for random coincidence noise and dead time, but not for tissue attenuation or scatter noise. Image reconstruction employed 2D Fourier rebinning followed by OSEM (4 iterations, 16 subsets), MAP (uniform resolution, smoothing parameter=0.01) or filtered backprojection (FBP; ramp filter, cutoff = Nyquist limit). Tumor image intensities were quantified as maximum single voxel value and as averages above thresholds = 90, 80 and 70% of the maximum. Images were calibrated for absolute AC by scanning a 24 cc plastic vial containing known concentrations of 18F or 64Cu. Results: Below 37 mg, there was a sharp decline in the ratio of image-derived to directly-measured AC, presumably reflecting a partial volume effect. Best overall accuracy and precision from 37-227 mg (n=16) were obtained using OSEM with maximum voxel [image/direct assay (mean±sd) =1.06±0.17]. MAP tended to overestimate AC (best result: 70% threshold, image/direct assay = 1.09±0.25), while FBP consistently underestimated AC (best result: max voxel, image/direct assay = 0.86±0.13). Conclusions: The results suggest that, in mice, phantom-calibrated small-animal PET is capable of measuring activity concentrations of superficial tumors in the 40-230 mg range with limited bias (<10%) and an acceptable coefficient of variation (<20%) using OSEM. ER -