Abstract 2147: Novel Procedure for Left Ventricular Myocardial Perfusion Quantification using Contrast Echocardiography: Clinical Studies and Validation in Animal Experiments
Exact myocardial perfusion imaging in contrast echocardiography (CE) requires tedious work, and is susceptible to interobserver variability. It often uses high energy flashes, which may cause premature ventricular contractions and microvascular leakage. We developed a novel modality to automatically evaluate regional myocardial perfusion using low mechanical-index (MI) pulses without flashes. In addition, regional wall motion is determined automatically based on the endocardial boundaries defined by the algorithm, which thus assesses perfusion and function simultaneously.
Methods: Fifteen patients with acute coronary syndromes were examined. Imaging was performed after myocardial contrast reached steady state. The myocardial location within each frame was automatically determined, and regional backscatter estimated after automatic compensation for signal attenuation. Blinded assessment of regional contraction was performed by 2 echocardiographers, and compared with the algorithm’s output. To validate outline detection, the same clinicians manually outlined the LV cavity at end-systole and end-diastole in ten loops, and these were compared with the algorithm. The algorithm was validated in 6 open chest sheep in 3 stages: baseline, reperfusion (15 minutes after 5 minutes occlusion of a coronary artery), and acute MI (15 minutes occlusion of the coronary artery). In each stage contrast-echo was performed, and perfusion assessed offline using our method on raw data obtained from one beat.
Results: Human studies: Seventeen cine-loops were available for analysis. The probability for perfusion levels in normokinetic and akinetic segments to belong to two identical groups (Wilcoxon Rank-Sum) was only 8310−7. The mean absolute error between the manually drawn outlines (8.2 ± 3.2) was very similar to the mean absolute error between each manually drawn outline and the automatically drawn outlines (7.0 ± 2.4, 8.8 ± 2.8).
Sheep studies: The sensitivity and specificity of our method in detecting segments affected by coronary artery occlusion were 0.833 and 0.854 respectively. The area under the ROC curve was 0.914.
Conclusion: Our innovative technique provides reliable simultaneous assessment of myocardial perfusion and function using CE.