Experimental methods
Six swine were induced with Pentothal 25 mg/kg, intubated, and
maintained on a respirator with halothane. Venous access was obtained at
3 jugular and two femoral sites. Two screw-in leads were inserted from
jugular sites, their positions captured on fluoroscopy and 3D mapping.
Leads were subsequently repositioned to different pacing sites,
simulating different focal arrhythmia sites with a minimal
inter-electrode distance of 10 mm. Screw-in leads were used for greater
stability and greater pacing consistency. A CS catheter was likewise
inserted via jugular access. Mapping was performed using a
high-resolution catheter with 8 spines and 48 electrodes (Octaray,
Biosense Webster, Irvine, CA) inserted via the femoral veins (Figure 1).
Additional pacing was performed via a mapping catheter in the femoral
vein.
For each swine, continuous pacing at the same cycle length was performed
at 2 to 3 different right atrial sites at least 10 mm apart for 60-90
seconds (Figure 1). In each animal, the electrodes were repositioned,
and a second set of maps was created. The operator was aware of the
change in pacing site and switched to the appropriate map manually.
Manually created maps were optimised to best represent each pacing site
through operator intervention. Using Carto Replay and retrospective
Parallel Mapping, the same mapping time was processed using ICPM as an
active filter, with the system automatically identifying the change in
pattern (alternating pacing site) and assigning each pattern to its
respective map, without operator intervention.
To validate ICPM performance and accuracy, each set of maps (manual and
automated) for each pacing site were compared offline using two methods.
Local activation times were compared with differences <10 msec
considered a match. Only paired points within 5 mm of one another were
included. Secondly, we compared interpolated color-coded activation maps
generated from all available information in order to assess similarity
between manual maps and those created by ICPM. For this comparison, both
maps were based on the same mesh anatomy. Manual and automatically
acquired maps were also qualitatively compared visually.