Tissue characterization of atherosclerotic plaques by intravascular ultrasound radiofrequency signal analysis: An in vitro study of human coronary arteries

Background Conventional gray-scale images of intravascular ultrasound (IVUS) cannot accurately differentiate histologic subtypes of sonolucent coronary plaques with or without a lipid core. Methods We analyzed radiofrequency signals obtained in vitro from 24 regions of interest (ROI) of noncalcified (sonolucent) plaques in 10 atherosclerotic coronary artery specimens pressure-fixed by formalin. Radiofrequency signals were sampled with a 30-MHz IVUS catheter and digitized at 500 MHz in 8-bit resolution. The ROIs were histologically categorized into 1 2 plaques with a lipid core and 12 plaques without it. integrated backscatter and statistical parameters of the radiofrequency envelope (mean/SD ratio [MSR], skewness, and kurtosis) within the ROI were calculated offline, and their ability to detect a lipid core was compared with visual analysis of the IVUS video images. In the group with lipid cores, percent area of a lipid core in each ROI was measured in a digitized histologic image by a computerized planimeter. Results Sensitivity and specificity of MSR, skewness, and kurtosis for lipid core detection were substantially greater than visual video image analysis (83.3% and 91.7%, 100% and 91.7%, 100% and 91.7% vs 53.3% and 71.7%). Furthermore, the parameters of integrated backscatter, MSR, skewness, and kurtosis were significantly correlated to percent of core area (r - -0.64, -0.73, 0.78, and 0.63, respectively; P < .05). Conclusions Compared with IVUS video images, the parameters of radiofrequency signal analysis may be used to aid in more accurate detection and quantitative evaluation of a lipid core, which is one of the major factors of a vulnerable coronary plaque.