Catheter Ablation -> Ventricular Arrhythmias -> Mapping & Imaging D-PO06 - Poster Session VI (ID 26) Poster

D-PO06-093 - Accuracy Of Non-invasive Localization Of Low-voltage Zones Using A Commercial Ecg-imaging System (ID 1361)

Abstract

Background: Non-invasive localization of low-voltage zones (LVZ) identifying myocardial scar would improve risk assessment and treatment in patients with ventricular arrhythmias.
Objective: Assess accuracy of non-invasive ECG Imaging (ECGI) for LVZ localization.
Methods: 18 VT patients underwent ECGI and invasive electroanatomical mapping (EAM) simultaneously. After EAM-ECGI co-registration, each virtual ECGI electrode site was paired with the closest EAM point within 20 mm and labelled as LVZ or non-LVZ based on EAM bipolar voltage (with both 0.5 and 1.5 mV used as thresholds). The following indices were automatically measured at each ECGI site: Unipolar (UA) and bipolar amplitude (BA), number of negative deflections during the QRS (ND), activation and activation-recovery interval gradients (GAT and GARI), minimum of the first derivative (dVdtMIN).
Results: 18 EAM and ECGI maps were compared (8 EPI, 8 LV and 2 RV). 14,178 ECGI and EAM sites were paired (median 698, interquartile range 562-1155 across patients). Of these, 54.2% (64.1%, 38.5-75.9%) and 30% (33.5%, 10.6-42.3%) were labelled as V<1.5 and V<0.5 mV, respectively. Pooling data together, LVZ showed smaller UA, BA, and dVdtMIN and higher ND (P< Bonferroni’s corrected threshold), but similar GAT and GARI (Figure 1). However, binary classification between sites with V<0.5 vs V>0.5 or between V<1.5 vs V>1.5 was poor, with areas under the ROC curve ≤0.56 and best accuracy ≤62% (Figure 1). Patient by patient analysis confirmed poor accuracy of ECGI indices for LVZ localization, with similar results for epicardial and endocardial EAMs.
Conclusion: Non-invasive discrimination of LVZ based on ECGI is not accurate.

Collapse