Influence of random, pile-up and scatter corrections in the quantification properties of small-animal PET scanners

The potential of PET imaging for pre-clinical studies will be fully realized only if repeatable, reliable and accurate quantitative analysis can be performed. The characteristic blurring of PET images due to positron range and non co-linearity, as well as random, pile-up and scatter contributions, that may be significant for fully 3D PET acquisitions of small animal, make it difficult their quantitative analysis. In this work specific activity versus specific counts in the image calibration curves for 3D-OSEM reconstructions from a commercially available small animal PET scanner are determined. Both linear and non-linear calibration curves are compared and the effect of corrections for random and scatter contributions are studied. To assess the improvement in the calibration procedure when scatter and random corrections are considered, actual data from a rat tumor pre- and post-cancer therapy are analyzed. The results show that correcting for random and scatter corrections can increase the sensitivity of PET images to changes in the biological response of tumors by more than 15%, compared to uncorrected reconstructions.