REGIONAL SPATIAL CLASSIFICATION FOR ATHEROSCLEROTIC PLAQUE CHARACTERIZATION WITH ULTRASOUND IMAGING

摘要

Objective. Ultrasound (US) characterization of the carotid atheroma may effectively describe plaques at a high risk of embolization. Clinical analysis has been based on visual inspection of two-dimensional (2D) US images or statistics of US echo amplitude histograms. We developed a novel US echo analysis technique for classification of regional areas within the plaque and tested its feasibility in vitro. Methods. US images of 12 carotid endarterectomy specimens were obtained at 7-10 MHz (high frequency, HF) and 3-6 MHz (low frequency, LF). After stretching gray scale to the 0-255 range, US echo amplitudes were classified into one of 10 regions (10R) within the plaque image. Percent area for each region was estimated. Based on echogenicity, Plaque Heterogeneity and Plaque Lucency Indices (PHI, PLI) were calculated for each plaque and imaging frequency. Results. PHI varied from 0.83 to 0.95. Average PHI were 0.89±0.03 for LF10R and 0.91±0.03 for HF10R. The difference between PHI for LF10R and HF10R was significant by paired t-test (p = 0.007). PLI varied from 27% to 79%. Average PLI were 55% ± 15% for LF10R and 53% ± 13% for HF10R. The difference between PLI for LF10R and HF10R was not statistically significant (p = 0.31). Conclusions. A technique for regional classification of carotid atheromatous plaques was successfully implemented. Simple indices for characterizing heterogeneity and echolucency quantified differences among carotid endarterectomy plaques. Heterogeneity was higher for the higher US frequency but echolucency was not.