A recovery coefficient method for partial volume correction of PET images

Correction of the "partial volume effect" has been an area of great interest in the recent times in quantitative PET imaging and has been mainly studied with count recovery models based upon phantoms that incorporate hot spheres in a cold background. The goal of this research study was to establish a similar model that is closer to a biological imaging environment, namely hot spheres/lesions in a warm background and to apply this model in a small cohort of patients. A NEMA phantom with six spheres (diameters 1-3.7 cm) was filled with (18)FDG to give sphere:background activity ratios of 8:1, 6:1, and 4:1 for three different acquisitions on a Philips Allegro scanner. The hot sphere SUVmax and the background average SUV were measured for calculation of recovery coefficients (RCs). Using the RCs, the lesion diameters, and the lesion:background ratio, the SUVmax of 64 lesions from 17 patients with biopsy proven lung cancer were corrected. The RCs versus sphere diameters produced characteristic logarithmic curves for each phantom (RCs ranged from 80% to 11%). From a cohort of 17 patients with biopsy proven lung cancer, 64 lesions combined had a mean SUVmax of 7.0 and size of 2.5 cm. After partial volume correction of the SUVmax of each lesion, the average SUVmax increased to 15.5. Hot spheres in a warm background more closely resemble the actual imaging situation in a living subject when compared to hot spheres in a cold background. This method could facilitate generation of equipment specific recovery coefficients for partial volume correction. The clinical implications for the increased accuracy in SUV determination are certainly of potential value in oncologic imaging.