Page 133 - YORAM RUDY BOOK FINAL

P. 133
4. Early Study Using Tikhonov Zero Order Regularization (Ramanathan, Nature Medicine

2004;10:22) 278 . Non-simultaneous RV pacing and LV pacing in the same patient. Site of
earliest activation, as determined from ECGI reconstructed epicardial potential maps,
corresponded well with the pacing lead terminal determined from CT for both RV and
LV pacing, as did the sequence of epicardial activation. ECGI located the pacing sites to

within 7mm (RV) and 11mm (LV). The published figure shows translucent views of the heart;
allowing the reader to see the pacing lead and potential map simultaneously for comparison
of pacing site location.

Atrial Pacing:

Atrial Fibrillation ECGI Study (Cuculich, Circulation, 2010;122:1364) 279 . In the testing
(validation) phase of this study, pacing was performed with a catheter in 6 patients from 11

different sites, known to be AF initiation sites. A total of 37 paced events. Location of pacing site
was determined directly by electro-anatomic mapping with CARTO. ECGI determined pacing
locations were accurate within 6.3 ± 3.9 mm.

Comparison to Intraoperative Mapping in Patients (Ghanem, Heart Rhythm, 2005;2:339) 280 .

Three patients undergoing open-heart surgery were studied during sinus rhythm and right
ventricular endocardial and epicardial pacing (total of five datasets). Body surface potentials were

acquired preoperatively or postoperatively using a 224-electrode vest. Heart-torso geometry was
determined preoperatively using computed tomography. Intraoperative mapping was performed
with two 100-electrode epicardial patches. A limitation of this study was the nonsimultaneous
acquisition of the surgical and noninvasive ECGI data under different conditions (open vs closed

chest). Because invasive and noninvasive data were not obtained simultaneously, time alignment
of the two cardiac sequences was necessary. Noninvasive potential maps captured epicardial
breakthrough sites and reflected general activation and repolarization patterns, localized
pacing sites to ~ 1 cm and distinguished between epicardial and endocardial origin of activation.

Noninvasively reconstructed electrogram morphologies correlated moderately with their invasive
counterparts (cross correlation = 0.72 ± 0.25 [sinus rhythm], 0.67 ± 0.23 [endocardial pacing], 0.71 ±
0.21 [epicardial pacing]). Noninvasive isochrones captured the sites of earliest activation, areas of
slow conduction, and the general excitation pattern.

A later intraoperative study (Zhang, J Am Col Cardiol (JACC) EP 2017;3:894–904; online
supplement) 280a focused on validation of repolarization parameters. The morphology of the
noninvasively reconstructed EGMs was very similar to that of the measured EGMs. About 82% of

reconstructed electrograms had correlation coefficients CC > 0.9 when contaminated by signal
noises and geometry errors. ECGI also reconstructed accurately the ARI patterns, identifying the
regions with longest and shortest ARI. ARIs were reproduced with good accuracy; 78% of
reconstructed electrograms had ARI that differed from the measured ARI by less than 10 ms.