New semiquantitative parameters in digital [18F]FDG PET/CT improve diagnostic accuracy in suspected infective endocarditis.

Background The purpose of this study was to identify semiquantitative parameters of [18F]FDG PET/CT using a digital PET scanner, which may increase diagnostic accuracy and readers' confidence in the diagnosis of infective endocarditis (IE). Results Images of 82 patients undergoing [18F]FDG PET/CT for suspected IE were visually and semiquantitatively analyzed. Standardized uptake values (SUV) of suspected foci, also normalized to liver, mediastinum and surrounding activity were calculated. For each, best thresholds were identified to diagnose endocarditis. Final diagnosis was reached by consensus in a multidisciplinary board. Semiquantitative analysis (SUVmax/SUVmax mediastinum, SUVmax/SUVmax liver, SUVpeak/SUVpeak mediastinum, SUVpeak/SUVpeak liver, SUVmax/SUVmean liver, SUVmax/SUVmean mediastinum, SUVmax focus/SUVmean focus, SUVpeak/SUVpeak surrounding) added to visual interpretation increased sensitivity (57-86%), specificity (83-93%), PPV (64-86%), NPV (79-93%) and diagnostic accuracy (74-90%) when using best SUVs thresholds (all p < 0.05). Conclusions Combining visual and semiquantitative analysis allows for a more accurate diagnosis of IE, and might be implemented into clinical routine.