Background—A method capable of determining atherosclerotic plaque composition and measuring plaque viscoelasticity can provide valuable insight into intrinsic features associated with plaque rupture and can enable the identification of high-risk lesions. In this article, we describe a new optical technique, laser speckle imaging (LSI), that measures an index of plaque viscoelasticity. We evaluate the potential of LSI for characterizing atherosclerotic plaque. Methods and Results—Time-varying helium-neon laser speckle images were acquired from 118 aortic plaque specimens from 14 human cadavers under static and deforming conditions (0 to 200 &mgr;m/s). Temporal fluctuations in the speckle patterns were quantified by exponential fitting of the normalized cross-correlation of sequential frames in each image series of speckle patterns to obtain the exponential decay time constant, &tgr;. The decorrelation time constants of thin-cap fibroatheromas (TCFA) (&tgr;=47.5±19.2 ms) were significantly lower than those of other atherosclerotic lesions (P<0.001), and the sensitivity and specificity of the LSI technique for identifying TCFAs were >90%. Speckle decorrelation time constants demonstrated strong correlation with histological measurements of plaque collagen (R=0.73, P<0.0001), fibrous cap thickness (R=0.87, P<0.0001), and necrotic core area (R=−0.81, P<0.0001). Under deforming conditions (10 to 200 &mgr;m/s), &tgr; correlated well with cap thickness in necrotic core fibroatheromas (P>0.05). Conclusions—The measurement of speckle decorrelation time constant from laser speckle images provides an index of plaque viscoelasticity and facilitates the characterization of plaque type. Our results demonstrate that LSI is a highly sensitive technique for characterizing plaque and identifying thin-cap fibroatheromas.

We describe laser speckle analysis for atherosclerotic plaque characterization and fibrous cap thickness estimation. Speckle analysis had a high sensitivity and specificity for detecting vulnerable atherosclerotic plaque, even in the presence of physiological motion.

A framework for the expression and analysis of statically defined communication patterns present in computations targeted for execution in the NuMesh multicomputer environment is developed. A system which solves optimization problems to statically allocate network resources for network traffic with real time constraints has been implemented. Results from the system, which uses simulated annealing and linear programming techniques to automate the process of placing and routing this traffic are presented. Thesis Supervisor: Stephen A. Ward Title: Professor of Electrical Engineering and Computer Science