Impact of Time-of-Flight and Point-Spread-Function in SUV Quantification for Oncological PET

BACKGROUND
Accuracy in the quantification of the SUV is a critical point in PET because proper quantification of tumor uptake is essential for therapy monitoring and prognosis evaluation. Recent advances such as time-of-flight (TOF) and point-spread-function (PSF) reconstructions have dramatically improved detectability. However, first experiences with these techniques have shown a consistent tendency to measure markedly high SUV values, bewildering nuclear medicine physicians and referring clinicians.

PURPOSE
We investigated different reconstruction and quantification procedures to determine the optimum protocol for an accurate SUV quantification in last generation PET scanners.

METHODS
Both phantom and patient images were evaluated. A complete set of experiments was performed using a body phantom containing 6 spheres with different background levels and contrasts. Whole-body FDG PET/CT of 20 patients with breast and lung cancer was evaluated. One hundred five foci were identified by 2 experienced nuclear medicine physicians.Each acquisition was reconstructed both with classical and advanced (TOF, PSF) reconstruction techniques. Each sphere and each in vivo lesion was quantified with different parameters as follows: SUVmax, SUVmean, and SUV50 (mean within a 50% isocontour).

RESULTS
This study has confirmed that quantification with SUVmax produces important overestimation of metabolism in new generation PET scanners. This is a relevant result because, currently, SUVmax is the standard parameter for quantification. SUV50 has been shown as the best alternative, especially when applied to images reconstructed with PSF + TOF.

CONCLUSIONS
SUV50 provides accurate quantification and should replace SUVmax in PET tomographs incorporating advanced reconstruction techniques. PSF + TOF reconstruction is the optimum for both detection and accurate quantification.

CITA DEL ARTÍCULO  Clin Nucl Med. 2013 Feb;38(2):103-9. doi: 10.1097/RLU.0b013e318279b9df