Effect of color coding and subtraction on the accuracy of contrast echocardiography

Background: Contrast echocardiography may be used to assess myocardial perf usion. However, gray scale assessment of myocardial contrast echocardiograp hy (MCE) is difficult because of variations in regional backscatter intensi ty, difficulties in distinguishing varying shades of gray, and artifacts or attenuation. We sought to determine whether the assessment of rest myocard ial perfusion by MCE could be improved with subtraction and color coding. M ethods and results: MCE was performed in 31 patients with previous myocardi al infarction with a 2(nd) generation agent (NC100100, Nycomed AS), using h armonic triggered or continuous imaging and gain settings were kept constan t throughout the study. Digitized images were post processed by subtraction of baseline from contrast data and colorized to reflect the intensity of m yocardial contrast. Gray scale MCE alone, MCE images combined with baseline and subtracted colorized images were scored independently using a 16 segme nt model. The presence and severity of myocardial contrast abnormalities we re compared with perfusion defined by rest MIBI-SPECT. Segments that were n ot visualized by continuous (17%) or triggered imaging (14%) after color pr ocessing were excluded from further analysis. The specificity of gray scale MCE alone (56%) or MCE combined with baseline 2D (47%) was significantly e nhanced by subtraction and color coding (76%, p<0.001) of triggered images. The accuracy of the gray scale approaches (respectively 52% and 47%) was i ncreased to 70% (p<0.001). Similarly, for continuous images, the specificit y of gray scale MCE with and without baseline comparison was 23% and 42% re spectively, compared with 60% after post processing (p<0.001). The accuracy of colorized images (59%) was also significantly greater than gray scale M CE (43% and 29%, p<0.001). The sensitivity of MCE for both acquisitions was not altered by subtraction. Conclusion: Post-processing with subtraction a nd color coding significantly improves the accuracy and specificity of MCE for detection of perfusion defects. (C) 1999 Elsevier Science Ireland Ltd, All rights reserved.