ECG Imaging to Detect the Site of Ventricular Ischemia Using Torso Electrodes: A Computational Study

Figure 1. Workflow for performing the forward problem. Transfer matrix (A) constructed using a 6,923-node male torso geometry and 2,499-node ventricle geometry, xt and yt are matrices filled with the epicardial ventricular potentials and torso potentials at each time t, respectively. Ventricular electrical excitations were simulated using a biophysically accurate 3D computer model of the human ventricles (Ni, 2016). The electrical action potentials on the epicardial surface are extracted and mapped onto the ventricular mesh representing the epicardial surface. Body surface potentials on the torso are then computed using the boundary element method (Stenroos and Haueisen, 2008).

Figure 2. Workflow for the inverse problem. Multi-lead ECG's using the various electrode configurations to the torso potential results from the forward problem. The BSP for the torso is then interpolated from the multi-lead ECG's. This interpolated BSP is then used as the torso potentials in the inverse solution to produce the epicardial potentials on the ventricle.

Figure 3. (A) Mean RDMS against lead layout, (B) 128 lead layout RDMS vs. timestep for the 3 Tikhonov solutions.

Figure 4. Snapshots showing the comparison between real and reconstructed epicardial excitation pattern. Both anterior and right lateral view of the ventricle were shown for over a period of 400 ms. (A) Real epicardial potential patterns at different timings. (B) Reconstructed epicardial potential pattern at different timings by using 128 leads and the 2nd order Tikhonov regularisation method.

Figure 5. Reconstructed epicardial excitation pattern from the right lateral view of the ventricle at 100 ms. Each regularisation column has two sub columns, (Left) is the reconstructed epicardial potential and (Right) is a “traffic light” diagram coloured according the ischemia detection results. Green areas represent the successful identification of the ischemic lesion, yellow areas are ischemic which were not identified, and red areas indicate areas incorrectly identified as ischemic.

Figure 6. The traffic light data from the ischemic lesion detection visualised as a bar chart. (A) 0 order Tikhonov, (B) 1st order Tikhonov, and (C) 2nd order Tikhonov. Green bars represent the successful identification of the ischemic lesion (correctly ischemic—CI), yellow bars are ischemic which were not identified and shown as healthy (incorrectly healthy—IH) and red bars indicate areas incorrectly identified as ischemic (incorrectly ischemic—II). All area values are represented as a percentage of the real circular lesion. Please note that some bars go above the height of the y axis, bar numbers are present on each bar to aid with this fact.

Figure 7. 3D ventricle detection results for 1st and 2nd order Tikhonov solutions for 128, 256, and 512 leads. Green areas represent the successful identification of the ischemic lesion, yellow areas are ischemic which were not identified, and red areas indicate areas incorrectly identified as ischemic.

Figure 8. Inner wall lesion detection. (A) 1st order Tikhonov 1,024 leads (B) 2nd order Tikhonov 1,024 leads.

Figure 9. Twelve lead ECG using BSP produced with the human ventricle-torso model. Simulation using healthy control conditions (black) and 5 ischemic lesion locations (5 colours). The y axis has been normalised between 1 and −1.

Figure 10. Variable location ischemic lesion detection results using traffic light representation. (A) Location vs. number of leads diagram showing the “traffic light” representation of the ischemic lesion detection results, a head on lesion view is shown for each location. First column shows the location of the ischemic lesion on the ventricle, remaining columns show a head on lesion view of the results (B) A bar chart visualizing the “traffic light” data taken from the whole geometry as views in (A) only show a single view. Please note that some bars go above the height of the y axis, bar numbers are present on each bar to aid with this fact.

Figure 11. Noise and sensitivity analysis on the lesion detection results and Tikhonov regularisation performance results including 20 dB SNR. (A) Signal to Noise ratio (10, 20, 30, 40 dB) vs. lesion threshold algorithm sensitivity with the lesion located on the left ventricle anterior wall. (B) Detection results using the 3 different regularisation techniques constructed from BSP's containing noise at 20 dB SNR using 256 electrodes layout. (C) Bar chart visualizing the “traffic light” data taken from (A) over the whole geometry. Bar height represents data taken from the Q = 0.4 in Equation (7) with error bars on the bar showing results of setting Q from 0.3 to 0.5. (D) Bar chart visualizing the “traffic light” data taken from (B) over the whole geometry.

Figure 12. (A–C) Three different lesion sizes reconstructed with the 2nd order Tikhonov solution and 512 leads. (D) Bar chart showing the areas incorrectly identified as ischemic in each solution as a percentage of the 27mm lesion area for comparison. Right lateral wall view and lesions have radius (A) 27 mm, (B) 20 mm, (C) 14 mm.