SUV Harmonization Between Different Hybrid PET/CT Systems

Current PET/CT lutetium oxyorthosilicate (LSO) or lutetium-yttrium oxyorthosilicate crystal scanners are equipped with sophisticated softwares including point-spread function (PSF) and time-of-flight (TOF) image reconstruction. These softwares are associated with increased SUVs compared with 3D-OSEM reconstructions associated with older BGO PET/CT scanners. The European Association of Nuclear Medicine (EANM) identified the problem of SUV harmonization since 2010 through the EANM Research Ltd European Association of Nuclear Medicine Research Ltd [EARL] FDG PET/CT accreditation program. This required processing 2 reconstructions, one optimized for maximum spatial lesion detection and one for harmonized quantitation. We investigated an alternative single reconstruction method for both qualitative and quantitative analysis optimized to maximize spatial lesion detectability, followed by an intrinsic postreconstruction algorithm for SUV harmonization.

Phantom and "in vivo" patient data analysis were acquired and analyzed on (a) a Siemens Biograph mCT system with LSO crystals and PSF and TOF algorithms, and on (b) a General Electric Discovery STE system with BGO crystals, without PSF and TOF. A dedicated algorithm (EQ.filter) was tested to harmonize SUV between the 2 scanners compared with EANM/EARL specifications. NEMA IQ phantom and a Jaszczak cylindrical phantom equipped with small fillable spheres (lesion to background ratios of 8:1 and 4:1) were used. Phantom data were validated on 7 oncologic patients with 39 hyperactive lesions ranging from 3 mm to 26 mm.

The main benefit of PSF + TOF LSO PET/CT systems was increased contrast for small active lesions. Recovery coefficients measured according to NEMA standards exceeded those obtained by 3D-OSEM reconstruction. SUVmax discrepancies between the 2 PET/CT systems were as high as 149%, dropping to below 10% when optimized by EQ.filter.

A single reconstruction optimized by EQ.filter for maximum spatial lesion detectability is an easy and precise solution to harmonize SUVs between different PET/CT scanners, avoiding a second reconstruction with an additional smoothing filter as requested by EANM/EARL.