REPOLARIZATION CHANGES DISPLAYED IN SURFACE ARI MAPS.
A SIMULATION STUDY
M. Tysler, M.Turzova, J.Svehlikova
Institute of Measurement
Science, Slovak Academy of Sciences
Abstract:
Body surface distribution of activation-recovery
intervals (ARI) and its ability to reflect local variations of the repolarization
process in different myocardium segments was analyzed. Body surface potentials were simulated for
normal activation - repolarization and in cases with local shortening and/or
decrease of the action potential (AP) of myocardial cells. Isotropic model of
analytically shaped ventricles and cellular automata were used to simulate the
spread of activation. AP was changed in defined regions in the anterior and
posterior left ventricle (LV) that represented 3 to 12% of the ventricular volume.
Corresponding body surface integral maps and ARI maps were analyzed. While shortening
of AP in the anterior LV was clearly projected on the left antero-lateral superior
torso, AP changes in the posterior LV projected mainly in the middle part of
the posterior torso were influenced also by other
processes in the myocardium and hardly to distinguish. Obtained results indicate
that ARI maps may reflect local changes of repolarization in both subendocardial
and transmural myocardial regions and could help to identify such regions namely
if they are close to the anterior chest surface.
INTRODUCTION
Vulnerability to ventricular arrhythmias
is connected with inhomogeneity of myocardium repolarization caused by local
changes of action potential duration and amplitude.
Under simplified conditions, ARI measured as interval between
the most negative derivative within the QRS complex and the most positive derivative
near the T-peak of an electrographic signal can be considered as some projection
of AP duration in the underlying myocardium. Similarly, ARI estimated from many
surface ECG signals, though strongly influenced by the torso volume conductor,
could be used as some indicator of local repolarization properties. Tank experiments
showed that despite the smoothing effect of torso, there might be a high correlation
between epicardially recorded AP duration and superficially measured ARI. Preliminary
study of surface ARI maps from real measurements [1] confirmed their reproducibility
that was much better than that of QT intervals. The aim of this study was to
test the ability of surface ARI maps to reflect AP changes in the heart and
their localization.
METHOD
Finite element
model of heart ventricles [2] with analytically defined geometry and element
size of 1mm3 was used to represent myocardium depolarization and repolarization. Conduction
velocity and AP shape were defined for each element. Several layers of elements
with up to five different AP durations decreasing from endocardium to epicardium
were used to build up the ventricular walls and the septum. AP were approximated
by step upstroke, constant plateau and 90 ms linear down-slope. AP duration
measured in the middle of the down-slope was 126 to 162 ms in the RV and 138
to 177 ms in the LV and in the septum. A layer with 3 times increased conduction
velocity on the endocardial surface simulated Purkinje fibers. Starting points
of activation were in agreement with experimentally observed early-activated
regions in a normal human heart. Activation spread was governed by a cellular
automata supposing isotropic myocardial tissue. 168 segmental dipoles were used
to represent the cardiac electric generator. Potentials on the surface of a
realistic torso model with basic inhomogeneities representing lungs and heart
cavities were computed using the boundary element method. ECG signals from 84
points of a 12x7 mapping grid were used to obtain potential and integral maps
as well as surface isochronal ARI maps. Normal heart repolarization and repolarizations
with AP locally shortened by 25% and/or decreased by 30% from the normal values
were simulated. Regions of changed AP were defined in two positions in LV as
shown in Fig.1: anteriorly near the apex and postero-laterally close to the
heart base. In both positions, lesions of three different sizes were created
and represented 3-12% of the myocardial volume. Small and medium lesions were
subendocardial while the biggest one was always transmural.
a)
b)
Fig. 1: Regions with
changed AP in the left ventricle:
a) anterior region (3%, 6% and 10% of the volume)
b) posterior region (4%, 8% and 12% of the volume)
RESULTS
The overall patterns
of simulated normal body surface potential maps as well as the patterns of ARI
maps were in good agreement with those measured in real subjects [1]. Example
of simulated normal ARI map and ARI map corresponding to shortened AP duration
in the anterior region are shown in Fig.2. Based on the definition, ARI were
evaluated only for ECG signals with positive T wave while ARI in areas with
negative T waves (upper right anterior and posterior torso) were not considered
and are not displayed.
TABLE I
Comparison of normal and changed ARI and integral maps for changed repolarization
in anterior and posterior heart regions
Figure 2. Example of simulated ARI maps. Left part
of each maps represents anterior torso, right part the back.
ARI durations in milliseconds are represented by
gray levels. ARI map for normal AP (upper) and for shortened AP in transmural
anterior region 10% of volume (lower map).
Changes of AP
simulated in anterior regions were projected mainly to the left antero-lateral
superior torso (in the middle of the map, near to the transversal level) but
partially also to the left inferior posterior torso. AP changes in posterior
LV regions close to the heart base were projected mainly in the middle part
of the posterior torso and partially also to the left lateral superior torso.
While merely decrease
of AP was difficult to recognize in the ARI maps, shortening of AP was clearly
visible and changes in the map were proportional to the size of the lesion.
Combined shortening and decrease of AP strengthened the changes in ARI maps.
Findings in QRST integral maps were partially in contrast with results in ARI
maps. AP decrease was reflected stronger than AP shortening and in most cases,
subendocardial lesions of the same size as transmural lesions produced greater
departures from normal QRST integral maps. Evaluation of changes in ARI maps
and QRST integral maps for medium and large lesions is in Table I a), b).
DISCUSSION
As it is difficult unambiguously interpret ARI for ECG tracings
with negative T wave, ARIs were not evaluated in this area (right superior part
of the torso). In a transition area between negative and positive T values,
T waves are small and often multi-phasic. In such region even small changes
of AP caused that the maximum of T wave derivative "jumped" to other
part of the T wave and possibly different processes in the myocardium were mixed
in the computed ARI.
Limitation of the study is the use of isotropic myocardium
model. As reported elsewhere, this can cause inaccuracy of the simulated potentials
namely if the regions with changed AP were not transmural. Another limitation
arises from the simulated linear AP down-slope that can influence estimation
of the "recovery time instant".
Although the resolution of surface mapping is in principle
limited by the smoothing effect of torso, results of our simulations suggest
that AP shortening alone as well as in combination with AP decrease can be recognized
in surface ARI maps, particularly in regions underlying the anterior chest.
AP changes in posterior LV were also clearly visible in the middle of the inferior
posterior torso but the computed long ARI values mostly did not reflect “true”
AP changes but were influenced by projections from other parts of the myocardium,
probably including the left septum.
Acknowledgments: This work was supported by grant
2/1135/21 from the VEGA grant agency.
REFERENCES
[1] M. Tyšler, M. Turzová, S. Filipová "Spatial
distribution of QT-intervals in body surface potential maps from limited leads,"
in: Electrocardiology 2000, 2001, pp.149-154.
[2] V. Szathmáry, I. Ruttkay-Nedecký:
"Model study of effects of different repolarization patterns in the left
and right ventricle on the resultant cardiac vectors" in: Electrocardiology
2000, 2001, pp. 97-102.