Identifying locations of re-entrant drivers from patient-specific distribution of fibrosis in the left atrium

Figures

• 

  Fig 1. Workflow for identifying patient-specific target areas with the highest probability of harbouring RDs. The personalized left atrial models are generated by segmentation of patient-specific LA geometry and fibrotic regions from LGE-MRI scans.

• 

  Fig 2. Segmentation and modelling of fibrosis. Labelling of voxels in the LA patient geometries according to their LGE MRI intensity ratio (IIR) relative to the blood pool. Voxels with IIR>1.24 are considered to be part of a fibrotic core. IIR<1.08 corresponds to healthy tissue and the intermediate values of IIR form a fibrotic border zone with intermediate properties.

• 

  Fig 3. Patient-specific fibrosis distribution in 6 AF patients with Utah scores from 2 to 4. The fibrotic regions are colour-coded to show dense fibrotic tissue (red) surrounded by a BZ of intermediate properties. The healthy atrium is shown in white. LIPV: left Inferior PV, RIPV: Right Inferior PV, LSPV: left Superior PV, RSPV: Right Superior PV and LAA: Left Atrial Appendage.

• 

  Fig 4. Regions of RD stabilisation in patient-specific LA models. The image shows bar chart with percentage of RDs found in different LA regions (blue: fibrosis, orange: PVs and grey: healthy LA tissue) after 6s of the simulation in the 6 patient-specific LA models with (A) and without fibrosis (B). A) In Utah 2 patients, the primary location for the RDs is the PVs. In Utah 3 patients, RDs are distributed between fibrotic regions and the PVs. In Utah 4 patients with severe fibrosis, the primary location of RDs are the fibrotic regions. FB: fibrotic burden.

• 

  Fig 5. CA strategies tested using virtual ablation platform. (A) Strategy 1: Circumferential PVI, (B) Strategy 2: PVI plus additional linear lesions at the LA roof and a line joining the PV to the MV, (C) Strategy 3: ablation on the TAs, (D) Strategy 4: TA guided ablation with additional linear lesions leading to unexcitable boundaries (PV and MV) and (E) Strategy 5: TA guided ablation with linear lesions (Strategy 4) plus PVI (Strategy 1).

• 

  Fig 6. The effect of fibrosis on the RD dynamics in a patient-specific LA model. A & C: the colour-coded voltage maps (normalised, so that 0 is rest state and 1 is maximal activation) in the LA of Patient 3 over 6s of simulation with the yellow dot indicating the RD tip and the arrow pointing to the direction of rotation. B & D: the corresponding RD trajectories (in black) superimposed on the fibrosis distribution. In simulations with fibrosis (A-B), the primary RD drifts around and across the fibrotic patch within the first 1s and thereafter stabilises at a distinctive location within the fibrotic BZ. However, in the same simulation without fibrosis (C-D), the primary RD remains at the location of its initiation.

• 

  Fig 7. The anchoring of RDs to fibrotic patches in patient-specific LA models. For all the 6 AF patients (rows), sections A and B show the AF simulation outcomes for 2 different RD initiation sites. In each section, on the left are the colour-coded voltage maps and on the right, the respective RD trajectories (black) superimposed on the fibrosis distributions. The yellow arrows indicate the direction in which the RD drifted over 6s of the simulation. The colour map used in the figure is the same as Fig 6.

• 

  Fig 8. Patient-specific RD location maps–potential CA targets. The images show colour-coded fibrosis distributions in the 6 patient-specific LA models (A-F), with the TAs superimposed. In Utah 3 and 4 categories (Patients 1, 2 and 3), TAs are seen at specific locations within fibrotic patches and at their BZs. In Utah 2 category (Patients 5 and 6), TAs are seen on the LA wall near small patches. Additionally, the TAs identified from simulations performed without fibrosis in the same LA model are presented in S3 Fig.

• 

  Fig 9. Comparison of the fibrosis burden to TA volume of in each patient-specific LA models. A: distribution of TAs across different regions of the atrial wall; dense fibrotic region (black), border zone (BZ—dark grey) and healthy tissue (light grey). The majority of TAs lie in the fibrotic region in Patients 1, 2 and 3 (FB > 25%, Utah 4 and 3), compared to much lower numbers in Patients from 4, 5 and 6 (FB < 25%, Utah 3 and 4). B: bar chart with FB (blue) and volume of all identified TAs (orange) in each model. The difference between FB and TA volume decreases with the decrease in Utah score, showing that the predicted TAs can be most efficient in improving CA in Utah 3 and 4 patients.

• 

  Fig 10. The outcome of CA using different strategies in Patient 2 (A) and Patient 3 (B). The summary of MF recorded for these cases are presented in the S1 Table.

• 

  Fig 11. CA with the clinical used ablation Strategies 1 and 2 failed to terminate AF in the majority of cases. Left: Voltage maps showing the behaviour of RDs after: (A1-B1) Strategy 2 and (C1-D1) Strategy 2 in Patient 2 and 3, respectively. Right: Transmembrane voltage at the point indicated by * in the respective LA models, plotted before (red) and after (blue) CA. Both strategies failed to terminate RDs in all cases and MF remained unaffected. Here and in the next figure, the white arrow indicates the directions of RD movement.

• 

  Fig 12. Target area guided ablation with Strategy 5 successfully terminated AF. Voltage maps showing the behaviour of RDs after CA of: (A1-B1) TAs and (C1-D1) TAs plus linear lesions joining them to the PVI lesions and the mitral valve (MV) in Patient 2 and 3, respectively. Panels on the right (A2-D2) show the voltage at the point indicated by * in the respective LA models, before (red) and after (blue) CA. Out of the two strategies, PVI plus TA guided ablation successfully terminated AF in both patients.