A computational investigation into rate-dependant vectorcardiogram changes due to specific fibrosis patterns in non-ischæmic dilated cardiomyopathy

Figures

• 

  Fig. 1. The computational mesh used in this study. Top Left: Image of the whole torso mesh, with transparency added to enable viewing of the bones, lungs, kidneys, liver, spleen, stomach and major arteries. Top Middle: Example of the electrical potential on the surface of the skin at a point during ventricular depolarisation/repolarisation. Top Right: Slice through the torso at the level indicated by the dashed line on the middle image at the same time point as the centre image, showing the electrical potential distribution throughout the torso, including the heart. Bottom: ECG trace for the case of no scar, paced at a basic cycle length of 600 ms. Since only ventricular activation is simulated, only the QRS complex of the ECG is reproduced.

• 

  Fig. 2. Sizes of small (left) and large (right) scars in the LV free wall (top) and septum (bottom). The UVC parameters used to define scar sizes (ϕ: rotational, ρ: transmural and z: apicobasal) are illustrated on the top right image, corresponding to the largest LV scar, with arrows indicating the direction of parameter increase. ϕ ranges from 0 in the middle of the septum to π∕2 at the outer surface of the anterior left/right ventricular junction, to π in the LV free wall; the inverse scale occurs in the posterior LV. ρ ranges from 0 on the endocardial surface to 1 on the epicardial surface. z ranges from 0 at the apex to 1 at the base.

• 

  Fig. 3. Activation maps for control (left), LV free wall scar (middle) and septal scar (right), for BCL of 300 ms (top) and 600 ms (middle); the scars are the largest simulated. The difference between the activation maps for 600 ms and 300 ms is shown at the bottom. All activation maps for BCL =300ms, BCL =600ms and for the differences between the two share a colour-scale; these are shown to the right of the rows. In all cases, data are plotted on the epicardial surface, along with clipping planes used to highlight endocardial and midwall data.

• 

  Fig. 4. Orientation of mesh with respect to x, y and z axes far left. VCG traces for control (second from left), LV free wall scar (second from right) and septal scar (far right); the traces for fast pacing (BCL = 300 ms) is represented by a solid line, the trace for slow pacing (BCL = 600 ms) by a dashed line. In each group, the 3D representation of the VCG is shown in the top left corner, with the x-y plot shown top right, the x-z plot bottom left, and the y-z plot bottom right; all axes range −1.5→1.5.

• 

  Fig. 5. Changes in metrics between fast and slow pacing (left). Changes in the maximum dipole and QRS area are expressed as percentage changes from their value for BCL =600ms. The mean unit weighted dipoles for the VCG during QRS are shown (right); a solid line represents the value for BCL =300ms, a dashed line for BCL =600ms.

• 

  Fig. 6. Changes in metrics between fast and slow pacing versus volume and surface area of scar, for both septal and LV free wall scars. ΔVCGmax and ΔQRSarea are expressed as percentage changes from slow pacing, ΔQRSd is expressed as the difference between slow and fast pacing, and dTVCGmean, dTVCGmax and dTQRSend are expressed as angular differences between the dipole vectors.

• 

  Fig. 7. Changes in metrics between fast and slow pacing versus ϕ, ρ and z, for both septal and LV free wall scars. ΔVCGmax and ΔQRSarea are expressed as percentage changes from slow pacing, ΔQRSd is expressed as the difference between slow and fast pacing, and dTVCGmean, dTVCGmax and dTQRSend are expressed as angular differences between the dipole vectors.

• 

  Fig. 8. Effect of scar density on the changes in dipole maximum magnitude ΔVCGmax (top left), ΔQRSd (middle left), ΔQRSarea (bottom left), dTVCGmean (top right), dTVCGmax (middle right) and dTQRSend (bottom right).

• 

  Fig. 9. Relative importance of each of the measured metrics in the random forests to classify scar location as either LV or septum.