Back to 2016 Annual Meeting
Carotid Plaque Strain Characterization in Clinical B-mode Ultrasound Image Sequences
Amir A. Khan, PhD1, Joseph C. Hecker, MS1, Tom Hatsukami, MD2, Siddhartha Sikdar, PhD1, Brajesh K. Lal, MD3.
1George Mason University, Fairfax, VA, USA, 2University of Washington, Seattle, WA, USA, 3University of Maryland School of Medicine, Baltimore, MD, USA.
Current risk-stratification for patients with carotid disease is dependent on an imperfect indicator, namely percent stenosis. Hypoechoic plaques may indicate a high risk of disruption and stroke. In addition, hemodynamic forces from pulsatile blood flow result in repetitive displacement and deformation (strain) of the plaque. We hypothesized that a) clinical ultrasound-imaging can be utilized to reliably measure carotid plaque strain characteristics and b) hypoechoic plaques would demonstrate increased strain.
B-mode video-loops of carotid plaques were acquired in longitudinal view using a standard linear-array transducer. 22 plaques were imaged in asymptomatic patients with stenosis ≥50%. Imaging was repeated to determine reproducibility. Plaques were outlined in multiple frames and optical flow methods used to track movement of pixels between frames in peak-systole and end-diastole. Magnitude of displacement was expressed as Maximum Absolute Shear Strain Rate (MASSR), while variation in displacement within the plaque was expressed as two novel strain metrics: Entropy of Shear Strain Rate (ESSR) and Variance of Shear Strain Rate (VSSR). Plaque composition was expressed as the percent plaque-area occupied by hypoechoic (GSM90). Strain measurements were compared to plaque echogenicity. The second image-acquisition was used to assess reproducibility using Intra-Class Correlation (ICC) and Bland-Altman plots.
A subgroup (36%) of the asymptomatic patients demonstrated high strain (MASSR 2.29) and heterogeneous strain-patterns (ESSR 40.47, VSSR 0.027) in their plaques, compared to the normal group (MASSR 0.57, ESSR 19.05, VSSR 0.002). Plaques with abnormal strain metrics had significantly larger hypoechoic regions (32% vs 18% in plaques with normal-strain) and smaller hyperechoic regions (8.2% vs 14.6% in plaques with normal-strain). Bland-Altman plots demonstrated excellent reproducibility with an ICC of 0.78.
Ultrasound image sequences acquired as part of a routine clinical vascular exam can be used to estimate plaque strain measures, and identify plaques under large biomechanical shear strain. The protocol is highly reproducible. A subset of carotid plaques in asymptomatic patients demonstrate abnormal strain metrics and these plaques have more hypoechoic material. The combination of abnormal strain in a hypoechoic plaque may indicate a high risk for stroke. Prospective studies will be required to validate these novel findings.
Back to 2016 Annual Meeting