Revised non-contact mapping of ventricular scar in a post-infarct ovine model with validation using contact mapping and histology

Abstract

Identification of arrhythmogenic scar using non-contact (NC) sinus rhythm (SR) mapping is limited. Dynamic substrate mapping (DSM) overcomes these limitations but is less accurate than plunge needle electrode mapping. We developed a revised method for calculating DSM which was validated using detailed histological analysis and compared with conventional mapping modalities. Mapping was performed in eight sheep, >9 weeks post-myocardial infarction. Twenty multielectrode needles were deployed at thoracotomy in the left ventricle within and surrounding scar, and located using Ensite. Simultaneous catheter, needle, and NC electrograms were recorded during SR and multisite pacing. Dynamic substrate mapping maps were calculated as the maximum local peak negative voltage (PNV). Absolute mean DSM (AMDSM) maps, based on peak-peak voltage (P-PV), were calculated to minimize local pacing effects and take into account anisotropic influence. Dynamic substrate mapping and AMDSM maps were normalized based on global maximum voltages attained. Histologically quantified scar and mapping criteria were compared using Spearman's correlation and receiver operator curves (area under the curve, AUC) using 50% scar cut-off. For unipolar mapping, needles had greatest sensitivity at identifying scar which was better for P-PV (AUC; needle = 0.90, catheter = 0.70, NC = 0.66) than for PNV (AUC; needle = 0.79, NC = 0.38). AMDSM (AUC = 0.75) had superior scar discrimination than either catheter (AUC; unipolar = 0.70, bipolar = 0.71) or DSM (AUC = 0.67). Absolute mean DSM accuracy was improved when valvular geometries were excluded (AUC = 0.77). Absolute mean DSM was comparably accurate in identifying scarred myocardium as PNV needle mapping but was superior to conventional catheter and NC mapping.