PVE correction in PET-CT whole-body oncological studies from PVE-affected images images

Partial Volume Effect (PVE) in PET-CT oncological studies affects the estimation of quantitative parameters useful for lesion malignancy differentiation and for monitoring disease and response to therapy. The aim of this work was to investigate the clinical feasibility and accuracy of PVE correction methods based on Recovery Coefficients (RC) as function of measured Lesion-to-Background ratio (L/B) _m and measured lesion volume (V _m). PET-CT measurements were performed. The radioactivity concentration (C _m) and V _m were measured from images using Operator-Dependent and Operator-Independent (OI) techniques. RC curves were obtained combining RC from NEMA 2001 IQ phantom measurements as function of Sphere-to-Background ratio (S/B) _m and sphere V _m. PVE correction was applied to PET-CT studies of anthropomorphic oncological phantoms and to PET-CT oncological studies (basal and follow up). The underestimation of C _m due to PVE in the NEMA IQ spheres (up to 80%) confirmed the severity of the error. The more feasible (always applicable, noise insensitive, reproducible) way to measure radioactivity was found by the use of an OI threshold technique. Our results showed that this measurement technique allows to achieve a PVE correction accuracy > 87% (error in radioactivity estimate <13%) for a sphere diameter > 1 cm. In patient studies the PVE correction was found to modify both SUV and SUV variations during patient follow up and our analysis showed that a PVE corrected SUV quantification increases the diagnostic confidence of oncological PET-CT studies.