High resolution, 3D isotropic late gadolinium enhanced imaging for the quantification of left atrial fibrosis and post-ablation scarring

Figures

• 

  Figure 1. (A) The left atrium (LA) was manually segmented on late gadolinium enhanced (LGE) images, and blood pool normalized image intensities (IIR) were projected onto a 3D surface model. Pulmonary veins and the mitral valve were excluded from analysis (excluded areas shown in grey). Fibrosis (defined as IIR values >1.34 based on a healthy control group’s mean plus 2 standard deviations) is shown in red, non-fibrotic tissue (IIR <1.34) in purple. (B) Bipolar electroanatomic mapping (EAM) was performed prior to pulmonary vein isolation (red dots indicate ablation sites). (C) EAM and LGE datasets were co-registered using rigid and non-rigid surface registration. (D) Quantification of LA fibrosis was performed 1) for the total LA surface 2) regional for the posterior wall only 3) on a point-by-point basis and 4) within peri-ablational regions in comparison to the remote myocardium.

• 

  Figure 2. Image intensity ratio (IIR) distribution in healthy controls (green), paroxysmal AF (lilac), and persistent AF (red) patients. The dashed line indicates the fibrosis threshold (IIR = 1.34), defined as the mean plus two standard deviations (SD) of the healthy control group’s IIR values.

• 

  Figure 3. Total fibrosis burden, expressed as the number of surface cells with an IIR > 1.34, in healthy controls (green), paroxysmal AF (lilac), and persistent AF (red) patients. Statistical significance was assessed using Dunn’s test with Bonferroni correction; bars and asterisks indicate significant differences (*P < 0.05; ns = not significant).

• 

  Figure 4. Correlation between total fibrosis burden on late gadolinium enhancement (LGE)—expressed as the percentage of surface cells with an image intensity ratio (IIR) > 1.34—and the percentage of low-voltage areas (LVA defined as regions with voltage < 0.5 mV) on bipolar electroanatomical mapping (EAM). No significant correlation was observed on a per-patient basis for the total left atrial surface (A) or the posterior wall (B).

• 

  Figure 5. Point-by-point comparison of late gadolinium enhancement (LGE) and bipolar electroanatomical mapping (EAM) data after rigid and non-rigid registration for (A) the total left atrial (LA) surface and (B) the posterior LA wall. Only points within a maximum distance of 5 mm were included in the analysis. Colour code: black = healthy tissue (IIR < 1.34, voltage > 0.5 mV); yellow = fibrosis by LGE only (IIR > 1.34, voltage > 0.5 mV); turquoise = fibrosis by EAM only (IIR < 1.34, voltage < 0.5 mV); red = fibrosis by both modalities (IIR > 1.34, voltage < 0.5 mV). The y-axis is log-scaled for visualization purposes.

• 

  Figure 6. Study flow chart; AF, atrial fibrillation; CMR, cardiovascular magnetic resonance; PM, pacemaker; PVI, pulmonary vein isolation.

• 

  Figure 7. (A) Late gadolinium enhancement (LGE) images of the left atrium before (at baseline) and six months after pulmonary vein isolation (PVI). Yellow arrows indicate dense ablation induced scarring at follow-up. The corresponding 3D models show LGE quantification in the peri-ablational area (±7.5 mm from ablation sites) in comparison with the remote atrial wall (turquoise). Pulmonary veins and the mitral valve were excluded from remote myocardium analysis. Areas with IIR > 1.34 (scar) appear red; IIR < 1.34 (healthy myocardium) appear lilac. (B) Boxplots of total LGE extent before and after PVI demonstrate significant fibrosis increase in peri-ablational regions, with no significant change in remote myocardium. Significance was tested by paired Wilcoxon test (*P < 0.05; **P < 0.01; ***P < 0.001).