Combined spatiotemporal and frequency-dependent shear wave elastography enables detection of vulnerable carotid plaques as validated by MRI

Figures

• 

  Figure 1. Overview of methodological setup. Ultrasound B-mode and SWE imaging is performed, providing both spatiotemporal group velocity (time-domain) and frequency-dependent phase velocity (after Fourier transform, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\boldsymbol{ {\mathcal F} }}$$\end{document}ℱ). Velocities are validated and correlated with MRI-derived plaque components, as well as AHA type classification. For the ultrasound and SWE panels, the plaque is delineated in blue. In the MR-vessel wall imaging (VWI) panel, one of the 2D black-blood VWI series is displayed with the analysis from MRI PlaqueView superimposed including the lipid-rich necrotic core (LRNC, yellow), intraplaque hemorrhage (IPH, red), calcifications (CA, white), loose matrix (purple), and vessel outline (blue). In the 3D volume rendering (VR) MR angiography (MRA) + VWI panel, the detected plaque components are superimposed on the MRA and displayed as a volume rendered image using the same color scheme.

• 

  Figure 2. Overview of generated SWE images. From left to right: Ultrasound B-mode images, SWE acquisition, ultrafast motion images obtained from data autocorrelation (four snapshot motion images displayed, from upper left to lower right), axial velocity map (space-time domain) with TTP-estimated group velocity (red slope), and Fourier-generated dispersion behavior and phase velocity map (velocity-frequency domain). Examples are given in longitudinal and transverse view for one AHA type VI, and one AHA type V plaque, respectively. Note that for the dispersion curves, data was discarded in frequency-ranges where corruption from noise or higher-order modes seemed visually apparent (in this case, lower frequency-range data for the AHA type V plaque in the longitudinal view plan was discarded from further analysis, as was higher frequency-range data for the AHA type VI plaque in transverse view). For all other presented cases, the 200–500 Hz range was deemed appropriate for inclusion. All examples are shown over a frequency range of 0–750 Hz.

• 

  Figure 3. SWE velocities, classified according to AHA type. Data sorted in longitudinal (light blue) and transverse view (dark blue), shown as group velocity (top left), and phase velocity at 200–300 Hz (top right), 300–400 Hz (bottom left) and 400–500 Hz (bottom right), respectively. Significance against all other groups is given as *(p < 0.05) or **(p < 0.01).

• 

  Figure 4. Selected significant correlations with group velocity. Data shown for longitudinal (left column) and transverse view (right column). All acquisitions for a given plaque is represented by their mean value (black dots), with an added Pearson linear regression line (red dashed). Significant correlations (*) are given for |R| > 0.50 and p < 0.05.

• 

  Figure 5. Selected significant correlations with phase velocity. Data shown for longitudinal (left column) and transverse view (right column). All acquisitions for a given plaque is represented by their mean value (black dots), with an added Pearson linear regression line (red dashed). Significant correlations (*) are given for |R| > 0.5 and p < 0.05.

• 

  Figure 6. Comparison between group velocity derived using the approach described in the method section, vs. using a vendor-specific system-derived shear wave output. (a) Correlation plot in longitudinal view, with a linear regression slope of m = 0.55. (b) Correlation plot in transverse view, with a linear regression slope of m = 0.89. (c) Vendor-specific system-derived group velocity as a function of AHA type, where no statistical differentiation could be inferred for the AHA type VI plaques against all other evaluated plaque types, respectively.