7 - Evaluation of LGE-CMR derived conducting channels for predicting VT ablation sites in nonischemic cardiomyopathy – direct comparison with electroanatomic maps

Background: LGE CMR enables accurate characterization of fibrosis in patients of nonischemic cardiomyopathy. Electroanatomic mapping (EAM) is used to locate areas in fibrotic tissue giving rise to ventricular tachycardia (VT) and guide ablation. However, there are limitations with voltage mapping such as errors in identifying arrhythmogenic substrate, limited spatial resolution as well as long procedure time. Applying a pixel signal intensity (PSI) algorithm to LGE-CMR can generate maps to instantaneously locate conducting channels (CC) in fibrotic tissue which form VT substrate.  A conducting channel (CC) in a nonischemic scar is a region of heterogeneous LGE within dense LGE which may correspond to abnormal voltage potentials within the arrhythmogenic substrate. We aimed to establish PSI thresholds to derive CC from LGE CMR in patients of NICM and then compare LGE CMR derived CC sites with EAM after actual co-registration of LGE-CMR PSI maps with EAM maps for diagnostic accuracy in predicting sites of VT ablation.

Methods: We evaluated retrospectively 45 patients with nonischemic cardiomyopathy and VT who underwent CMR prior to EAM and VT ablation.  Short-axis LGE acquisitions were processed using ADAS-VT (Galgo Medical SL, Barcelona) to generate CC within myocardial layers by a PSI-based algorithm characterizing fibrosis as dense or heterogeneous scar at 60-40% PSI. This data was co-registered with EAM on Carto-system and evaluated as per AHA-17-segment model for ablation sites on a per patient and per channel basis. The PSI thresholds were then adjusted for each patient to obtain a best match with EAM abnormalities and compared with the 60-40% PSI thresholds.

Results: In 45 patients, EAM-CMR merge with adjusted PSI-thresholds for dense-scar at 57.93% ± 9.1% of maximum PSI and heterogeneous-scar at 39.17% ± 7.95%, identified 274 potential VT sites out of which 96 were ablation targets. 137 CC-sites were identified on CMR.. The overall diagnostic accuracy of this technique improved from 69.1% to 76.92 % at the adjusted thresholds. Further, 55% fractionated potential and 61% pacing-sites were accurately identified by the CC-method.

Conclusion: LGE-CMR derived CC showed high sensitivity and negative predictive value and a good diagnostic accuracy in identifying EAM sites of ablation. Varying the PSI thresholds of dense and heterogeneous scar on LGE-CMR in individual patient further improved the diagnostic accuracy of the technique. Thus target sites of VT ablation can be predicted from PSI maps derived from LGE-CMR in patients of nonischemic cardiomyopathies. It can also predict potential sites of recurrent VT not targeted at the time of EAM but corresponding to late potentials. q