Methods for the computational localization of atrio-ventricular pre-excitation syndromes

The site of atrioventricular pre-excitation can roughly be estimated with the help of schemes basing on a few number of electrocardiogram (ECG) leads. Computer algorithms have been developed which utilize the body surface mapping of the pre-excitation signal for the localization purpose. We tested several new algorithms.

A patient suffering from Wolff-Parkinson-White syndrome was investigated prior the catheter ablation. The body surface mapping was performed with a 62-lead magnetocardiograph. The site of pre-excitation was calculated by using different methods: the dipole method with fixed and moving dipoles, the dipole scan on the endocardium, and different current density methods (L1 norm method, L2 norm method, low resolution electromagnetic tomography (LORETA) method, and maximum entropy method). Three-dimensional (3D) magnetic resonance imagings (MRIs) of the heart were used to visualize the results. The source positions were compared to the site of catheter ablation.

The accessory pathway was successfully ablated left laterally. This site was correctly identified by the conventional dipole method. By scanning the entire endocardial surface of the heart with the dipole method we found a circumscribed source area. This area too, was located at the lateral segment of the atrio-ventricular grove. The current density methods performed differently. Whereas the L1 norm identified the site of pre-excitation, the L2 norm, the LORETA method and the maximum entropy method resulted in extended source areas and therefore were not suited for the localization purpose.

The dipole scan and the L1 norm current density method seem to be useful additions in the computational localization of pre-excitation syndromes. In our single case study they confirmed the localization results obtained with the dipole method, and they estimated the size of the suspected source region.