Page 127 - YORAM RUDY BOOK FINAL

P. 127
The GMRes method 262 is an iterative scheme that does not apply constraints. As such, it is

an independent approach, unrelated to TR. It does not require a priori knowledge of the solution
properties (for imposing appropriate constraints). The overall accuracy of GMRes and TR ECGI
solutions is very similar. However, in certain cases GMRes reconstructs localized epicardial
potential features (e.g., multiple potential minima) that are lost in TR due to the smoothing effect

of the imposed constraint. As such, GMRes is well suited for reconstructing localized events such
as multiple epicardial breakthrough sites or the origin of an arrhythmia. Application of the two
complementary methods (TR and GMRes) in clinical ECGI ensures reliability and maximizes the
extraction of diagnostic information in the clinical setting.

MFS is a meshless method that does not require meshing (discretization) of the heart
261
and torso surfaces. As such, it differs from the BEM approach used in TR and GMres, that requires
meshing the surfaces. MFS employs a system of virtual (fictitious) sources to compute Φ . Its
E
accuracy is similar to that of BEM in the ECGI application . It has several advantages in the
261
clinical setting. The application is fast, because the meshing and mesh – optimization procedures
are time consuming and constitute the time-limiting step in ECGI; there is no need for mesh
optimization (a procedure that involves manual editing), which enhances automation of ECGI in

the clinical application; mesh-induced artifacts are eliminated, especially for the complex
geometries of the atria. Therefore, MFS is well suited for mapping atrial rhythms. Note that each
approach has its unique properties and the choice of method depends on the application.

Equation (5.6) is time-independent; it computes Φ at one instant of time. By repeating the
E
computation at consecutive time instants (typically 1 ms apart) one can obtain the entire set of
epicardial potential maps during an entire cardiac cycle (or many cycles). By plotting the potential
at one epicardial site over time, one obtains a local epicardial electrogram (EGM). This process can

be repeated for many points on the epicardial potential map (we typically compute 1000 EGMs).
Local activation time (AT) is determined by the steepest negative deflection – (dV/dt) max on the
QRS of the local EGM. Activation isochrones, depicting the excitation front and the sequence of
activation, are obtained by connecting all sites with the same activation time. Similarly, recovery

(repolarization) time (RT) is determined by the maximum positive deflection (dV/dt) max on the EGM
T-wave. The local activation-recovery interval (ARI) is then given by

ARI = RT – AT

ARI has been shown to be a surrogate for the local action potential duration (APD) 263,264 .

ECGI requires two sets of data: Φ , the potential on the entire torso surface, and matrix A
T
that contains the heart-torso geometrical relationship. Φ is recorded with many (250) electrodes
T
imbedded in a vest or strips and controlled by a computerized mapping system. Matrix A is
obtained using CT or MRI scan (typically with axial resolution between 0.6 and 1.0 mm) gated to
the ECG. Figure 5.1 shows a schematic diagram of the ECGI procedure.