Published in:
Open Access
01-12-2015 | Original Article
Harmonizing FDG PET quantification while maintaining optimal lesion detection: prospective multicentre validation in 517 oncology patients
Authors:
Elske Quak, Pierre-Yves Le Roux, Michael S. Hofman, Philippe Robin, David Bourhis, Jason Callahan, David Binns, Cédric Desmonts, Pierre-Yves Salaun, Rodney J. Hicks, Nicolas Aide
Published in:
European Journal of Nuclear Medicine and Molecular Imaging
|
Issue 13/2015
Login to get access
Abstract
Purpose
Point-spread function (PSF) or PSF + time-of-flight (TOF) reconstruction may improve lesion detection in oncologic PET, but can alter quantitation resulting in variable standardized uptake values (SUVs) between different PET systems. This study aims to validate a proprietary software tool (EQ.PET) to harmonize SUVs across different PET systems independent of the reconstruction algorithm used.
Methods
NEMA NU2 phantom data were used to calculate the appropriate filter for each PSF or PSF+TOF reconstruction from three different PET systems, in order to obtain EANM compliant recovery coefficients. PET data from 517 oncology patients were reconstructed with a PSF or PSF+TOF reconstruction for optimal tumour detection and an ordered subset expectation maximization (OSEM3D) reconstruction known to fulfil EANM guidelines. Post-reconstruction, the proprietary filter was applied to the PSF or PSF+TOF data (PSFEQ or PSF+TOFEQ). SUVs for PSF or PSF+TOF and PSFEQ or PSF+TOFEQ were compared to SUVs for the OSEM3D reconstruction. The impact of potential confounders on the EQ.PET methodology including lesion and patient characteristics was studied, as was the adherence to imaging guidelines.
Results
For the 1380 tumour lesions studied, Bland-Altman analysis showed a mean ratio between PSF or PSF+TOF and OSEM3D of 1.46 (95 %CI: 0.86–2.06) and 1.23 (95 %CI: 0.95–1.51) for SUVmax and SUVpeak, respectively. Application of the proprietary filter improved these ratios to 1.02 (95 %CI: 0.88–1.16) and 1.04 (95 %CI: 0.92–1.17) for SUVmax and SUVpeak, respectively. The influence of the different confounding factors studied (lesion size, location, radial offset and patient’s BMI) was less than 5 %. Adherence to the European Association of Nuclear Medicine (EANM) guidelines for tumour imaging was good.
Conclusion
These data indicate that it is not necessary to sacrifice the superior lesion detection and image quality achieved by newer reconstruction techniques in the quest for harmonizing quantitative comparability between PET systems.