Article Text


149 Automated analysis of atrial ablation-scar using delayed-enhanced cardiac MRI
  1. L Malcolme-Lawes1,
  2. R Karim1,
  3. C Juli2,
  4. P B Lim2,
  5. T V Salukhe2,
  6. D W Davies2,
  7. D Rueckert1,
  8. N S Peters1,
  9. P Kanagaratnam2
  1. 1Imperial College London, London, UK
  2. 2Imperial College Healthcare, London, UK


Introduction Visualisation of the ablation-related atrial scar using delayed-enhanced MRI (DE-MRI) may reveal important underlying causes for atrial fibrillation (AF) recurrence following ablation. In order to develop and objective method for delineating ablation-scar we compared pre and post DE-MRI after Cryo-balloon lesion on the basis that a more predictable lesion set would be created for validation.

Methods and Results 12 patients undergoing cryoablation for PAF were enrolled in the study, and underwent pre-ablation DE-MRI scans. Pulmonary vein isolation (PVI) was confirmed in all patients at the end of the cryoablation procedure using a circular mapping catheter. Additional ablation by RF or Freezer Max was required to achieve PVI in 59%. No ablation was performed in any region other than the PV ostia. Post-ablation DE-MRI was performed at 3 months. An automatic segmentation of the LA was produced with custom software from the MRA sequence. The preablation and postablation free breathing late gadolinium enhanced sequence was registered to the MRA and the maximum intensity within the LA wall was projected onto the post ablation LA surface. The blood pool was identified automatically using custom software as the region 1 cm inside the wall of the LA, and its mean (BPM) and SD used as a baseline. To identify a universal threshold for scar, regions of brightest myocardium were initially selected in pre and post ablation MRIs. The brightest regions were 1.9±1.2 vs 8.7±3.1 SDs above the BPM in pre-and post-ablation MRIs respectively (p=0.001). A threshold of 5 SDs above the BPM was therefore programmed into our custom software to identify regions of scar for all patients. The ostial regions were defined as extending 1 cm both proximal and distal to the PV–LA junction, and selected manually for left and right sided veins prior to scar projection. (See Abstract 149 figure 1). The scar proportion within these regions was calculated using commercially available software ITK-SNAP. Total LA scar proportion was 0.2±0.02% vs 6.3±0.75% in pre and post ablation scans respectively. The increase in scar seen in the PV ostia was 24.6±1.38% compared with 2.6±1.28% in the rest of the LA (p=0.01) (See Abstract 149 figure 2).

Abstract 149 Figure 1

Comparison of pre-ablation and post-ablation %scar using fixed threshold.

Conclusion We have demonstrated the feasibility an objective, automated method of DE-MRI analysis of left atrial ablation-scar. This technique will now need to be validated against clinical outcomes.

  • Atrial fibrillation
  • catheter ablation
  • MRI

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.