PVE correction in PET from PVE affected images

Partial volume effect (PVE) in positron emission tomography (PET), as a consequence of the PET limited spatial resolution, causes underestimation of radiotracer uptake in oncological lesions and needs to be corrected. Aim of this work was to develop a method for PVE correction of PET-CT oncological studies, based on curves of recovery coefficients (RC) as a function of measured Lesion-to-Background ratio (L/B_m) and measured lesion volume (V_m). Measurements were performed on the PET-CT scanner Discovery STE (GE Medical Systems). Measured radioactivity concentration (C_m) and V_m were estimated using both Operator-Dependent (OD) and Operator-Independent (OI) measurement techniques. RC curves were obtained by NEMA 2001 IQ phantom measurements. PVE correction was applied to anthropomorphic phantoms miming different body districts and oncological lesions. PVE correction was also applied to 16 oncological patients (both basal and follow up studies). The underestimation of C_m (up to 80%) confirmed the severity of the error. Residual errors between GS and PVE corrected C_m showed that the more accurate way to measure radioactivity at a regional level is by the use of an OI threshold technique. The results on the accuracy of the PVE correction method showed that the method achieves an accuracy > 88% for lesion diameter > 1cm. In patient studies PVE correction was found to modify both SUV and SUV variations (¿SUV) during patient follow up, and our analysis showed that a PVE corrected SUV quantification increases the diagnostic confidence. Results proved that the proposed PVE correction method, combined with an OI threshold technique for the measurement of the radioactivity uptake, is feasible and accurate in a clinical setting.