Atherosclerotic plaque rupture potential is conferred by plaque composition and structure. We have previously shown in humans in vivo that carotid plaque components can be automatically delineated by a support vector machine (SVM) classifier considering normalized crosscorrelation (NCC)-derived measures of ARFI-induced displacement. We now extend our prior work by hypothesizing that classification is improved by using displacements derived using blind source separation (BSS). In 20 carotid plaques imaged in vivo in patients undergoing carotid endarterectomy (CEA) were imaged prior to extraction, and specimens were harvested after CEA for histological processing. ARFI displacement profiles were calculated from each of the first five principal components of the RF data and used as inputs to the SVM classifier. The classifier was evaluated by 5-fold cross-validation, with the histological samples acting as gold standards. From the output SVM likelihood matrices, ROC curves were calculated for separating collagen from calcium and lipid-rich necrotic core from intraplaque hemorrhage. For all examined plaques, inputting displacement profiles derived from the first four eigenvectors to the SVM classifier increased sensitivity and specificity over using NCCderived displacement profiles. These results suggest that using BSS-derived displacement profiles as inputs to the SVM classifier improves discrimination of carotid plaque components that are correlated to vulnerability for rupture.