Extended Data Figure 9 : Measurement of intramural phase singularity dynamics using 2D ultrasound imaging and tracking of a non-moving heart versus a fibrillating contracting heart in 3D ultrasound data and optical mapping.

From: Electromechanical vortex filaments during cardiac fibrillation

Extended Data Figure 9

a, Schematic of the experimental setup for simultaneous fluorescence and 2D ultrasound imaging in intact, Langendorff-perfused rabbit hearts. Left, ultrasound imaging (Vevo 2100, VisualSonics Inc.) with the cross-sectional echocardiographic imaging plane aligned tangentially inside the left ventricular wall (white line). Imaging plane facing fluorescence imaging camera (Extended Data Fig. 3d). Right, action potential, calcium and strain imaging setup for fluorescence imaging. b, Schematic of imaging configuration with action potential wave (phase) imaged on the surface of the heart and rotating mechanical pattern (phase) imaged inside the ultrasound cross-section inside the ventricular wall. The rendering shows a scroll wave (green, computer simulation). c, Optical mapping shows a counter-clockwise rotating action potential wave (AP, green) on the surface of a rabbit heart during ventricular fibrillation (Supplementary Video 9). The electrical vortex is associated with a counter-clockwise rotating pattern of contractile (red) and tensile (blue) rates of strain observed optically with fluorescence imaging on the surface. The mechanical deformation observed in B-mode ultrasound imaging in an imaging plane located beneath the imaged surface and aligned approximately parallel or co-planar to the epicardium shows a phase singularity, corresponding to a counter-clockwise rotating wave in mechanical deformation (Supplementary Video 9). The analysis of the electrical and mechanical vortices reveals co-existing phase singularities (indicated by white circles) on the surface and inside the ventricular wall. d, Tracking of the non-moving heart compared to a fibrillating contracting heart in 3D ultrasound and fluorescence imaging data. Distributions showing magnitudes of tracked displacement vectors resulting from motion tracking of the 3D ultrasound data (blue and red curves) and motion tracking during optical imaging (yellow curve). Influences by speckle noise or possible residual motion onto the motion tracking and motion analysis during ultrasound imaging are minimal in the Langendorff setup. Motion is not visible (Supplementary Video 13, left) when the heart does not contract. In comparison, the small contractions and deformations during ventricular fibrillation can clearly be observed (Supplementary Video 13, right) and can also be detected using motion-tracking algorithms (red curve, 3D ultrasound; yellow curve, optical mapping). The maximum of tracked displacements for the asystolic heart is approximately 0.15 mm. By contrast, during fibrillation, the maximum displacements are clearly shifted towards larger values with a maximum at 0.45 mm. The magnitudes of the tracked displacements during ventricular fibrillation (red curve) are confirmed by the optical measurement performed at the same time (yellow curve). Statistical analysis (two-sample Kolmogorov–Smirnov test) rejects the null hypothesis that the ultrasound-based measurements of tissue displacement during asystole and ventricular fibrillation are from the same continuous distribution at a 1% significance level. For the ultrasound data, the distributions were obtained from 274,444 voxels (asystole) and 165,558 voxels (fibrillation), for which the displacements were tracked in between consecutive volume frames (with Δt = 1 or Δt = 2 or Δt = 3) for the entire video sequence. Then only the maximal displacement in each voxel was stored and considered for the distributions to emphasize the influence of noise. Note in this context that in the asystolic case, tracking yields small displacement magnitudes due to measurement noise resulting presumably from both the algorithm and speckle noise. For a better comparison, the Δt or the temporal distance between volume frames between which motion was detected was adjusted. The asystolic data were imaged at volume rates of 62 volumes per second, whereas the fibrillation data was imaged at volume rates of 91 volumes per second. For the asystolic data (left), displacements were computed in between one and the second next volume frame (Δt = 2), resulting in an effective volume rate of approximately 30 volumes per second. However, the distribution (left) remained unchanged with Δt = 1, Δt = 2 or Δt = 3 frames, indicating that the tissue is static and does not exhibit motion. For the fibrillation data (right), displacements were computed in between one and the third next volume frame (Δt = 3), resulting also in an effective volume rate of approximately 30 volumes per second. With Δt = 1 or Δt = 2 frames, the distribution also remains clearly distinguishable from the non-moving distribution and retains displacement magnitudes well over 0.5 voxels. Note that in this particular fibrillation dataset, the overall strength of motion and deformation is relatively small compared to other datasets (for instance the one shown in Fig. 1). In the optical data, displacements were computed from a frame at time t to a frame at time t + 32 ms (Δt = 16 frames) to achieve a corresponding measurement of tissue displacements at imaging speeds of approximately 30 frames per second and to ensure comparability of the data. The data and the clearly distinguishable visual appearance of the non-moving and fibrillating heart (Supplementary Video 13) demonstrate that the spatiotemporal resolution of the 3D ultrasound imaging is sufficient to resolve elastomechanical deformation patterns during tachyarrhythmias.