Abstract
1760
Objectives Interpretation of F-18-fluorodeoxyglucose positron emission tomography (FDG PET) myocardial viability images can be suboptimal in patients with high residual blood-pool activity. This study evaluated the effect of an automated blood-background correction on the interpretation of myocardial viability and scar scores.
Methods FDG and N-13-ammonia perfusion PET images were analyzed from 25 patients (10 with diabetes). Automated analysis was performed to calculate the percent of left ventricle (LV) mismatch (hibernating myocardium) and match (fibrotic scar) scores using in-house FlowQuant software, with and without prior subtraction of the blood-background. Expert visual analysis of uptake defects was performed on a 0-4 scale using the AHA-standard 17-segments, and expressed as %LV. FlowQuant match and mismatch scores were compared to expert visual interpretation as the clinical reference standard.
Results Patients had LV ejection fraction of 17-33% and blood glucose values of 4-15 mmol/L. FDG blood-background levels varied from 8 to 53%, and automated FDG defect scores increased on average by 25% after background subtraction (p< 0.001). The corresponding perfusion-FDG mismatch and match scores also increased by 27% and 22% respectively (p< 0.001), reflecting a greater extent and severity of disease detected. Mismatch scores were more accurate with vs. without background subtraction (−11% vs. -29% bias; p< 0.001), whereas match scores (+37% vs. +9% bias; p< 0.001) were substantially higher compared to visual interpretation. The results suggest that automated contast-recovery may increase sensitivity for quantification of the total scar burden in the presence of residual blood background activity.
Conclusions Contrast recovery corrrection improved agreement for the assessment of hibernating myocardium using automated vs expert visual mismatch scores. Further validation studies are required to determine the effect of mismatch and scar scores on prognosis for use in clinical practice.