The purpose of this study was to compare the diagnostic efficacy of attenuation-corrected and nonattenuation-corrected whole-body 18F-fluorodeoxyglucose (FDG) PET images to determine an adequate method that can semiquantitatively evaluate nonattenuation-corrected images.
Methods: Whole-body PET studies were performed in 24 fasting patients with various tumors (lung cancers, n = 18; mediastinal tumors, n = 4; breast cancers, n = 2) 30-40 min after a bolus injection of 18F-FDG. Transmission scans followed emission data acquisition. Reconstructed attenuation-corrected and uncorrected images were displayed simultaneously and the relative FDG uptake in lesions and corresponding background areas was evaluated by the region of interest method. Both types of images were also compared with X-CT scans and conventional nuclear medicine scans for diagnostic efficacy.
Results: Attenuation-corrected and uncorrected images were found to be equally sensitive for detecting lesions. There was a strong linear correlation between lesion-to-background (L/B) ratios calculated on attenuation-corrected and uncorrected images (r = 0.98; p < 0.001). Significant differences in L/B ratios between attenuation-corrected and uncorrected images were present in only 6 of 55 lesions (11%). Standardized uptake ratios (SURs) in attenuation-uncorrected images did not correlate with SURs in attenuation-corrected images nor with L/B ratios in uncorrected images.
Conclusion: The efficacy of attenuation-uncorrected FDG PET images in evaluating tumors is similar to that using attenuation-corrected images. Uncorrected images provide not only clinically useful but also quantitative information equivalent to that provided by attenuation-corrected images. However the L/B ratio is the only available index that can be used for quantification of uncorrected images.