Diagnosis of myocardial viability by dual-head coincidence gamma camera fluorine-18 fluorodeoxyglucose positron emission tomography with and without non-uniform attenuation correction.

This study assessed a dual-head coincidence gamma camera (hybrid PET) equipped with single-photon transmission for myocardial fluorine-18 fluorodeoxyglucose (FDG) imaging by comparing this technique with conventional positron emission tomography (PET) using a dedicated ring PET scanner. Twenty-one patients were studied with dedicated FDG ring PET and FDG hybrid PET for evaluation of myocardial glucose metabolism, as well as technetium-99m tetrofosmin single-photon emission tomography (SPET) to estimate myocardial perfusion. All patients underwent transmitted attenuation correction using germanium-68 rod sources for ring PET and caesium-137 point sources for hybrid PET. Ring PET and hybrid PET emission scans were started 61+/-12 and 98+/-15 min, respectively, after administration of 154+/-31 MBq FDG. Attenuation-corrected images were reconstructed iteratively for ring PET and hybrid PET (ac-hybrid PET), and non-attenuation-corrected images for hybrid PET (non-ac-hybrid PET) only. Tracer distribution was analysed semiquantitatively using a volumetric vector sampling method dividing the left ventricular wall into 13 segments. FDG distribution in non-ac-hybrid PET and ring PET correlated with r=0.36 (P<0.0001), and in ac-hybrid PET and ring PET with r=0.79 (P<0.0001). Non-ac-hybrid PET significantly overestimated FDG uptake in the apical and supra-apical segments, and underestimated FDG uptake in the remaining segments, with the exception of one lateral segment. Ac-hybrid PET significantly overestimated FDG uptake in the apical segment, and underestimated FDG uptake in only three posteroseptal segments. A three-grade score was used to classify diagnosis of viability by FDG PET in 136 segments with reduced perfusion as assessed by SPET. Compared with ring PET, non-ac-hybrid PET showed concordant diagnoses in 80 segments (59%) and ac-hybrid PET in 101 segments (74%) (P<0.001). Agreement between ring PET and non-ac-hybrid PET was best in the basal lateral wall and in the apical-septal segment (80%-100%), and lowest in the apical, supra-apical and posteroseptal segments (41%-55%). Ac-hybrid PET showed highest agreement in the lateral wall (89%-100%), and lowest agreement in the apical and the basal septal segments (59%-67%). In conclusion, non-uniform attenuation correction with singles transmission significantly improves the diagnostic accuracy of myocardial dual-head gamma camera coincidence imaging with FDG. However, results equivalent to those obtained with ring PET cannot yet be attained, even if attenuation correction is applied. New rebinning algorithms for three-dimensional data may further improve the performance of ac-hybrid PET and should be evaluated in future studies.