Imaging the dynamics of cardiac fiber orientation in vivo using 3D Ultrasound Backscatter Tensor Imaging

Figure 1. 3D-BTI principle. (A) Tilted plane waves are emitted and the associated backscattered echoes are recorded. (B) Coherent compounding is applied in postprocessing to synthetically generate voxel-specific focal zones. (C) The 2D spatial coherence function is calculated for each voxel and (D) an elliptic fit is applied to determine the fibers orientation. (E) A vector representation is then achieved to display the fibers orientation in 3D.

Figure 2. 3D-BTI of one ex vivo porcine myocardial sample. (A) 3D grayscale ultrasound image of one myocardial sample. (B,C,D) Three examples of spatial coherence functions obtained for different depth locations in the sample. (E) Drawing of the heart associated to the ultrasound acquisition location. (F) Vector representation used for the visualization of the transmural variation of the fibers in the entire sample. (G) Schematic representation of the reference coordinates. Local axes u, v and w are defined by the circumferential and longitudinal directions on the epicardial surface. Fiber angle α, is positive when measured counterclockwise from the v-axis. (Figure adapted from Streeter et al.27).

Figure 3. 3D-BTI validation against histology. (A) Five histological slices (20X) and associated BTI slices of a left ventricle myocardial sample highlights the transmural angle variation of the fibers. (B) An example of the fiber angles variation assessed with histology and with 3D-BTI through the wall thickness (0% endocardium −100% Epicardium) for the same sample is displayed. (C) Bland-Altman plot of the transmural fiber angles estimated by 3D-BTI and histology (n = 5 myocardium samples).

Figure 4. In vivo 3D-BTI. 3D representation of fibers orientation in the left ventricle of an open-chest sheep at four different moments of the cardiac cycle (i.e. (A) Late Diastole, (B) Early Systole, (C) Late Systole, (D) Early Diastole).

Figure 5. Fiber orientation temporal variation. In vivo fiber angle variation in the left ventricular wall within a cardiac cycle as a function of depth and time. The color scale represents the value of the fiber angle superimposed onto a standard M-mode of the left ventricle. The associated ECG is displayed at the bottom of the figure.

Figure 6. Transthoracic imaging of myocardial fiber orientation in the human heart. The fiber orientations of the antero-septal LV wall is shown in end-diastole and end-systole.

Figure 7. Principle of spatial coherence estimation on a 2D matrix array probe. (A) An ultrasound wave is focused in a random media by applying delays in emission. (B) The backscattered echoes are recorded for each element of the 2D array. The autocorrelation of the rephased signals is calculated for each pair of elements as a function of distance between elements that composed the pair. (C) Represents the theoretical coherence function for a focused emission in a random media that we would obtain with a matrix probe of N x × N y elements.