Yokokawa Miki, Jung Dae Yon, Joseph Kim K, Hero Alfred O, Morady Fred, Bogun Frank
Division of Cardiovascular Medicine.
Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, Michigan.
Heart Rhythm. 2014 Nov;11(11):1966-73. doi: 10.1016/j.hrthm.2014.06.036. Epub 2014 Jun 30.
Twelve-lead electrocardiogram (ECG) criteria for epicardial ventricular tachycardia (VT) origins have been described. In patients with structural heart disease, the ability to predict an epicardial origin based on QRS morphology is limited and has been investigated only for limited regions in the heart.
The purpose of this study was to determine whether a computerized algorithm is able to accurately differentiate epicardial vs endocardial origins of ventricular arrhythmias.
Endocardial and epicardial pace-mapping were performed in 43 patients at 3277 sites. The 12-lead ECGs were digitized and analyzed using a mixture of gaussian model (MoG) to assess whether the algorithm was able to identify an epicardial vs endocardial origin of the paced rhythm. The MoG computerized algorithm was compared to algorithms published in prior reports.
The computerized algorithm correctly differentiated epicardial vs endocardial pacing sites for 80% of the sites compared to an accuracy of 42% to 66% of other described criteria. The accuracy was higher in patients without structural heart disease than in those with structural heart disease (94% vs 80%, P = .0004) and for right bundle branch block (82%) compared to left bundle branch block morphologies (79%, P = .001). Validation studies showed the accuracy for VT exit sites to be 84%.
A computerized algorithm was able to accurately differentiate the majority of epicardial vs endocardial pace-mapping sites. The algorithm is not region specific and performed best in patients without structural heart disease and with VTs having a right bundle branch block morphology.
