Download Three-Dimensional Cardiac Electrical Imaging From Intracavity

1454
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 8, AUGUST 2007
Three-Dimensional Cardiac Electrical Imaging From
Intracavity Recordings
Bin He*, Fellow, IEEE, Chenguang Liu, Student Member, IEEE, and Yingchun Zhang
Abstract—A novel approach is proposed to image 3-D cardiac
electrical activity from intracavity electrical recordings with the
aid of a catheter. The feasibility and performance were evaluated
by computer simulation studies, where a 3-D cellular-automaton
heart model and a finite-element thorax volume conductor model
were utilized. The finite-element method (FEM) was used to
simulate the intracavity recordings induced by a single-site and
dual-site pacing protocol. The 3-D ventricular activation sequences as well as the locations of the initial activation sites were
inversely estimated by minimizing the dissimilarity between the
intracavity potential “measurements” and the model-generated
intracavity potentials. Under single-site pacing, the relative error
(RE) between the true and estimated activation sequences was
0 03
0 01 and the localization error (LE) (of the initiation
site) was 1 88 0 92 mm, as averaged over 12 pacing trials when
considering 25 V additive measurement noise using 64 catheter
electrodes. Under dual-site pacing, the RE was 0 04 0 01
over 12 pacing trials and the LE over 24 initial pacing sites was
2 28 1 15 mm, when considering 25 V additive measurement
noise using 64 catheter electrodes. The proposed 3-D cardiac
electrical imaging approach using intracavity electrical recordings
was also tested under various simulated conditions and robust
inverse solutions obtained. The present promising simulation
results suggest the feasibility of obtaining 3-D information of
cardiac electrical activity from intracavity recordings. The application of this inverse method has the potential of enhancing
electrocardiographic mapping by catheters in electrophysiology
laboratories, aiding cardiac resynchronization therapy, and other
clinical applications.
Index Terms—Cardiac electrical imaging, cardiac electrical
tomography, catheter ablation, catheter mapping, electrocardiographic imaging, inverse problem.
I. INTRODUCTION
ENTRICULAR arrhythmias account for nearly 400 000
deaths per year in the United States alone [1]. These lethal
arrhythmias typically occur in patients with structural heart disease. Pharmacologic treatment of these arrhythmias has been
difficult because of significant pro-arrhythmic effects of many
anti-arrhythmic agents, so more invasive approaches are being
utilized [2]. Cardiac ablation techniques have been used in a
growing number of patients to ablate critical sites in the heart
that initiate ventricular tachycardia (VT) [3]. Cardiac VT ablation is a complex invasive procedure that requires considerable
time in the electrophysiology laboratory to locate the sites of
V
Manuscript received April 14, 2006; revised November 15, 2006. This work
was supported in part by the National Science Foundation (NSF) under Grant
BES-0411480. Asterisk indicates corresponding author.
*B. He is with the University of Minnesota, Department of Biomedical Engineering, 7-105 NHH, 312 Church Street SE, Minneapolis, MN 55455 USA
(e-mail: [email protected]).
C. Liu and Y. Zhang are with the University of Minnesota, Department of
Biomedical Engineering, Minneapolis, MN 55455 USA.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TBME.2007.891932
initiation of VT that will be ablated. This often involves prolonged periods of keeping patients in VT to do catheter-based
mapping.
In the past decade, important advancements have been made
to map and localize cardiac electrical activity from the endocardium using catheter technology, including 1) CARTO mapping using a nonfluorescent electro-anatomic catheter mapping
technique recording the electrical potential and the geometric
data of the endocardium over a series of sites [4], and 2) endocardial inverse solutions, where a cavitary noncontact multielectrode catheter-probe is used to explore the inverse reconstruction of endocardial potentials from the potential measurements made on the catheter probe [5]–[7]. To date, the efforts of
catheter-based mapping approaches have been focused on obtaining electrical potentials or other electrical properties over
the surface of endocardium [4]–[7].
While these endocardial mapping and inverse solutions show
promise of offering a minimally invasive means of localizing
and mapping cardiac electrical activity over the endocardial
surface, no attempt has been made, to our knowledge, to image
the 3-D cardiac electrical activity from the intracavity catheter
recordings. Clearly, the availability of a 3-D tomographic
imaging approach that could image cardiac electrical activity
throughout the 3-D myocardium will provide a much desired
functional imaging means to assess cardiac electrical activity
and provide an alternative guide to catheter ablation and other
clinical applications.
In the present study, we have proposed a new approach for
imaging 3-D cardiac electrical activity from intracavity electrical measurements, with the aid of the finite-element method
(FEM). The feasibility of this new method is tested by computer
simulations.
II. METHODS
A. Imaging Principles
The concept of the 3-D cardiac electrical imaging from
intracavity electrical recordings is to estimate the 3-D cardiac
electrical activity by minimizing the difference between the
recorded and model-generated intracavity electrical signals at
an instant or during a period [33]. Such signals may be recorded
by a recording catheter, or on the surface of the endocardium.
within the 3-D myMathematically, the inverse solution
ocardium can be expressed as (1)
0018-9294/$25.00 © 2007 IEEE
Authorized licensed use limited to: University of Minnesota. Downloaded on August 21, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
(1)
HE et al.: THREE-DIMENSIONAL CARDIAC ELECTRICAL IMAGING FROM INTRACAVITY RECORDINGS
where
is the vector of the electrical sources,
and
are the vectors of the recorded and model-generated
electrical signals (electrical potentials as tested in the present
simulation study) within the intracavity. , , and are
parameters to determine the specific procedures of an inverse
refers to the period during which the inverse
solution.
, the problem reduces to
imaging is performed. When
is a weighting matrix, and
is
instantaneous imaging.
can be obtained by solving the
regularization parameter.
to the
forward problem from the cardiac electrical sources
intracavity electrical signals. Thus, the proposed 3-D cardiac
electrical imaging from intracavity electrical signals consists of
two major steps: 1) the forward procedure to calculate the intracavity electrical signals from cardiac electrical sources; and 2)
the inverse procedure to estimate 3-D cardiac electrical activity
from the measured intracavity electrical signals. The proposed
approach is tested by computer simulation as described below.
B. Forward Modeling
was obIn the present study, the forward solution
tained by using the FEM with the aid of a cellular-automaton
heart model [8]–[10]. The heart model was constructed from
CT scans of a human subject and contains about 65 000 myocardial cell units, at which an action potential is assigned. The
anisotropic propagation of excitation in the ventricular myocardium is incorporated into this computer heart model [10]
according to literature [11]–[14]. The myocardial fiber orientations rotated counterclockwise over 120 from the outermost
) to the innermost layer (endocardium,
layer (epicardium,
) with identical increment between the consecutive layers.
All the units on the same myocardial layer of ventricles had
identical fiber orientations. The conduction velocity was 0.8
m/s along the fiber and 0.3 m/s transverse to the fiber. The
heart model was made up of about 65 000 myocardial cell
units spaced by 1.5 mm. Each myocardial unit’s parameters,
such as the pattern of action potential and the vector of local
fiber orientation, were set individually. We considered the
anisotropy of intracellular conductivities when calculating the
instantaneous current sources in the cellular automaton model.
The conductivity tensor at each cell unit was computed from
the local fiber orientation. The three orthogonal components of
the current source at each cell unit were respectively computed
as the product of the negative spatial gradient of instantaneous
transmembrane potential for each of the three directions and the
corresponding intracellular conductivity at the same direction.
Based on the bidomain theory, the extracellular electrical potential within the thorax can be solved from the following:
(2)
(3)
where and are the intracellular and interstitial conductivity
the transmembrane
tensor, the equivalent current density,
potential, and the outward unit normal to the surface , which
includes the body surface and the surface of the catheter. in
(3) represents the physical quantity of electrical current density,
while in (1) refers to the vector of electrical sources at a finite
1455
number of locations within the myocardium ( maybe current
density or maybe some other biophysical quantities).
Based on the CT images of a human subject, a finite-element (FE) model was built to simulate the realistic geometry
thorax volume conductor. The CT images were segmented,
edge-detected, and contoured for the torso, lungs, epicardial
and endocardial surfaces, respectively. The surface contours
were meshed by triangles to build a boundary-element (BE)
model. The BE model of a balloon catheter in an inflated status
was embedded into (located approximately in the center of) the
cavity of the left ventricle. The triangulated surface models were
then transformed to volume definition model by Rhinoceros
software (Robert McNeel & Assoc., WA) using Non-Uniform
Rational B-Splines (NURBS), which is an accurate mathematical description of 3-D geometry, in order to generate FE
meshes. The FE model was obtained by meshing the integrated
NURBS geometry model. The heart-torso-catheter FE model
was made up of 20 423 nodes and 118 660 first-order tetrahedral
elements. The tetrahedral elements within the myocardium and
the blood mass had finer resolution as compared to other areas.
The torso, lungs and blood mass were assumed to be isotropic
conductors. The anisotropy of cardiac tissue was incorporated
into the computer model as described above. The conductivities
of different tissues were set as follows in the FE model: torso
(0.20 S/m), lungs (0.08 S/m), blood mass (0.6 S/m), cardiac
tissue (interstitial: 0.6 S/m along the fiber orientation and 0.15
S/m transverse to the fiber orientation; intercellular: 0.3 S/m
along the fiber orientation and 0.075 S/m transverse to the
fiber orientation) [10]. The inside of the catheter was assumed
nonconductive. In the present study, we used a rigid geometry,
as the investigation is focused on ventricular activation which
occurs during the isovolumetric contraction.
The cardiac current sources were simulated by means of
a cellular-automaton heart model [8]–[10] via (3) and used to
evaluate the electrical potential field in the FE mode via (2).
Equation (2) was discretized into a system of linear equations
defined on the FE nodes by means of the FEM (see [15] for details). The linear equations were solved by the preconditioned
conjugate gradients method, a linear solver, to obtain the electrical potential at every FE node [15]. Sixty-four electrodes were
uniformly distributed over the catheter surface and the electrical
potentials at these electrodes were calculated according to a certain cardiac status (such as sinus rhythm or pacing).
C. Inverse Approach
Fig. 1 shows the schematic diagram of the inverse procedure. We adapted a heart-electrophysiological-model-based inverse algorithm, which we have previously developed based
on body surface potential maps (BSPMs) [9], [10], to solve
(1) in the present simulation. Note that using this inverse alis different than . A preliminary classification
gorithm,
system which approximately determines cardiac status based
on a priori knowledge and the measured intracavity potential
maps (ICPMs) was used to estimate the initial pattern of myocardial activation by means of an artificial neural network [8].
According to the output of the preliminary classification system,
the parameters of the heart computer model were initialized, and
the corresponding ICPM was calculated using the FEM. The
Authorized licensed use limited to: University of Minnesota. Downloaded on August 21, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
1456
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 8, AUGUST 2007
Fig. 1. The schematic diagram of the 3-D cardiac electrical imaging approach
from the intracavity recordings. ICPM: Intracavity potential map. BSPM: Body
surface potential map. See text for details.
model’s parameters were iteratively adjusted in an attempt to
minimize the dissimilarity between the measured and the heartmodel-generated ICPMs, employing a multiobjective nonlinear
optimization procedure. The following equation:
(4)
was solved with the aid of the Simplex Method [16], where 1)
is constructed with the average correlation coefficient
(CC) between the measured and simulated ICPMs from instant
to instant
of the cardiac excitation after detection of initial activation and 2)
is constructed with the CC between the integral of the measured ICPMs and the integral of
the simulated ICPMs. The is a parameter vector consisting of
heart model parameters to be optimized in the process.
Fig. 2. The forward and inverse simulation results. Panel (a) shows the forward-calculated ICPMs, BSPM and Lead-2 ECG during sinus rhythm, by means
of the finite-element volume conductor model. Panel (b) shows two typical examples of the 3-D cardiac electrical inverse solutions from the ICPM, during
single-site pacing (left) and dual-site pacing (right). The true and estimated activation sequences are shown by isochrones of one longitudinal section and six
transverse sections, respectively. The locations of the true and estimated pacing
sites are illustrated by yellow dots on the isochrone maps.
uncertainty and variation were introduced to test the robustness
of the proposed approach. The parameters were chosen based
on possible ranges of the values in an experimental setting.
The 3-D inverse solutions were quantitatively evaluated by
two measures: localization error (LE) and relative error (RE).
The LE is defined as the distance from the pacing site to the site
of initiation of corresponding activation. The RE is defined as
follows:
(5)
D. Simulation Protocol
Computer simulations using a pacing protocol were used
to evaluate the performance of the present ICPM-based 3-D
cardiac electrical imaging approach. Under such pacing protocol, the parameters consisted of the pacing sites. Twelve
sites were selected and paced individually from the following
regions throughout the ventricles: BA: basal-anterior; BRW:
basal-right-wall; BP: basal-posterior; BLW: basal-left-wall;
BS: basal-septum; MA: middle-anterior; MP: middle-posterior;
MLW: middle-left-wall; MS: middle-septum; AA: apical-anterior; AP: apical-posterior; AS: apical-septum.
For each pacing simulation, assuming the peak-peak value of
the intracavity potential is 5 mV, Gaussian white noise (GWN)
was added to the calculated
of standard deviation of 25ICPMs to simulate the noise-contaminated ICPM measurements, which served as the input of the inverse approach.
Random noise (average value of 1.2 mm) was also added to
the electrode positions to simulate the electrode position uncertainty. Effects of catheter position shift were also evaluated
by introducing 5-mm shift of the catheter. These parameter
where
is the number of the ventricular cell units,
is the
is the inversely
true activation time of the th cell unit and
estimated activation time of the th cell unit.
III. RESULTS
The feasibility of solving the forward problem by means of
the FEM is shown in Fig. 2(a), where the simulated ICPMs,
BSPM, and Einthoven lead-II ECG are shown during sinus
rhythm. In this simulation study, the catheter was placed inside
the blood cavity of the left ventricle.
A typical example of the present inverse solution during
single-site pacing is shown in Fig. 2(b) (left panel). The ICPMs
from
to
after the onset of pacing
were used to inversely estimate the location of the pacing site
and the ventricular activation sequence. The LE was assessed
by the distance from the localized site of origin of activation
to the true pacing site. The estimation error for the activation
Authorized licensed use limited to: University of Minnesota. Downloaded on August 21, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
HE et al.: THREE-DIMENSIONAL CARDIAC ELECTRICAL IMAGING FROM INTRACAVITY RECORDINGS
1457
TABLE I
LE OF THE PACING SITE, AND RE OF THE RECONSTRUCTION OF ACTIVATION SEQUENCE, FROM THE 64-ELECTRODES ICPMS DURING SINGLE SITE PACING
TABLE II
LE OF THE PACING SITE, AND RE OF THE RECONSTRUCTION OF ACTIVATION
SEQUENCE, FROM THE 32-ELECTRODES AND 128-ELECTRODES ICPMS DURING
SINGLE SITE PACING
sequence was assessed by RE between the true and estimated
activation sequences.
The simulation results during single-site pacing over 12
pacing sites are shown in Table I. The mean and standard devimm and
,
ation of the LE and RE were
respectively, when only additive measurement noise of 25was considered. When both additive measurement noise and
electrode position uncertainty with an averaged value of 1.2 mm
were considered, the mean and standard deviation of the LE and
mm and
, respectively. When
RE become
both additive measurement noise and catheter shift of 5 mm
were considered, the mean and standard deviation of the LE
mm and
, respectively.
and RE became
Effects of number of recording electrodes over the catheter
are shown in Table II, where 32 or 128 electrodes were uniformly distributed over the catheter surface. Only measurement
noise with standard deviation of 25- V was considered in this
case. The LE and RE became
mm and
for
32-electrodes, and
mm and
for 128 electrodes. Compared with the results using 64 electrodes, no significant effects were observed when using 32 or 128 electrodes,
considering other possible error sources, such as a possible shift
in position of the catheter (5-mm shift led to an averaged LE of
4.84 mm).
The performance of the ICPM-based inverse approach was
also evaluated by the dual-site pacing. Twelve pairs of myocardial cell units in a seven-layer myocardial region adjacent to
the atrial-ventricular ring were randomly selected to simulate
two localized regions of activation. The similar noise and uncertainty conditions were simulated as in the single pacing protocol. Fig. 2(b) (right panel) shows a typical example of the inverse solutions during dual-site pacing with GWN of 25being added to the forward-calculated ICPMs. Table III lists
the LE and RE for all 12 pairs of pacing sites. On average, the
, and the LE over
RE of the activation sequence were
mm, when only addi24 initial activation sites was
tive measurement noise with standard deviation of 25was
considered. When both additive measurement noise and electrode position uncertainty with an averaged value of 1.2 mm
were considered, the mean and standard deviation of the LE and
mm and
, respectively.
RE increased to
When both additive measurement noise and catheter shift of 5
mm were considered, the LE and RE were further increased to
mm and
, respectively.
Effects of number of recording electrodes over the catheter
are shown in Table IV during dual site pacing, where 32 and
128 electrodes were uniformly distributed over the catheter surface. Only measurement noise with standard deviation of 25was considered in this case. The LE and RE were
mm and
for 32-electrodes, and
mm and
for 128 electrodes.
IV. DISCUSSION
We have proposed a novel approach for 3-D cardiac electrical
imaging from intracavity electrical recordings [33]. With the
aid of catheter mapping of intracavity electrical signals at multiple sites simultaneously, the present approach promises to provide an important advancement over conventional approaches
[5]–[7] by offering the 3-D ability for localizing and imaging
of cardiac electrical activity. We used the FEM to solve the forward problem and a cellular-automaton heart model to simulate
ventricular activation. The feasibility of imaging cardiac activation sequence and localizing the site of initiation of activation
has been shown in computer simulations using single-site and
dual-site pacing protocols. The present promising simulation results (on average 2–3 mm LE for single-site pacing, and 2.6–3.3
mm LE for dual-site pacing, when there is no catheter position
shift; on average less than 5 mm LE when there is catheter position shift), suggest the potential clinical applications of this
Authorized licensed use limited to: University of Minnesota. Downloaded on August 21, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
1458
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 8, AUGUST 2007
TABLE III
LE OF THE PACING SITES, AND RE OF THE RECONSTRUCTION OF ACTIVATION SEQUENCE, FROM THE 64-ELECTRODES ICPMS DURING DUAL SITE PACING
TABLE IV
LE OF THE PACING SITES, AND RE OF THE RECONSTRUCTION OF ACTIVATION
SEQUENCE, FROM THE 32-ELECTRODES AND 128-ELECTRODES ICPMS DURING
DUAL SITE PACING
new approach to accurately localize site of origin of ventricular
activation from the widely used catheter procedure in a clinical
setting.
The largest localization and reconstruction errors occurred
when there is a shift in position of the catheter. For a 5-mm
shift of catheter position, the resulting averaged LEs are 4.84
and 4.53 mm, respectively, for single site and dual site pacing.
This data suggest that there is a need to develop a technique to
accurately determine the position and geometry of intracavity
catheter in real time for clinical applications. On the other hand,
the present simulation results also suggest that the intrinsic error
of the present 3-D cardiac imaging approach from ICPMs is low,
if there is no such systematic shift in probe position. Note that
while we have evaluated effects of measurement noise and geometry and sensor location uncertainty in the present simulation
study, we assumed that the forward transfer matrix represents
well the lead field relating the electrical sources within the myocardium to the electrode sensors. The realistic effect of such
possible distortion would have to be assessed in an experimental
setting in either an animal model (e.g., [17]) or in human subjects (e.g., [18]).
Note that only ICPM data from a period of ventricular depolarization (21 ms–48 ms) were used to solve the spatio-temporal
inverse problem and activation sequence throughout the ventricular activation of up to 171 ms was computed. This was made
possible because of using of the cellular-automaton heart model
in which the ventricular activation is encoded.
Compared with the BSPMs which are measured over the torso
surface, the ICPMs were simulated in the present study within
the left ventricle. While the right ventricular wall had a larger
source-to-sensor distance, the present simulation results indicated that the proposed approach can image electrical activity
from both the left ventricle and the right ventricle. This shall be
explained by the fact that electrode sensors record electrical potentials via volume conduction and there are no insulating materials separating the left ventricle from the right ventricle.
Compared to the body surface recordings, the intracavity
recordings have the advantage of high signal-to-noise ratio
(SNR), since the effect of environmental noise is smaller in the
cavity than on the body surface and the signal is larger within
the intracavity compared due to smaller distance from the cardiac sources to the sensors. The smoothing effect caused by the
volume conductor also has less effect on the intracavity potentials, due to the vicinity of catheter probe and the myocardium
[7]. Thus, the ICPM-based cardiac electrical imaging, as a minimally invasive procedure, may offer higher spatial resolution,
and maybe easier to be used in combination with the ablative
catheter or catheters for other purposes in a single procedure.
On the other hand, the BSPM-based cardiac electrical imaging
is a complete noninvasive procedure, although the smearing
effect of the torso volume conductor may be larger than the
ICPM-based results. Future investigations should be made to
compare the ICPM-based and BSPM-based imaging results.
As the catheter mapping is widely practiced in a clinical setting to aid catheter ablation, the ICPM-based endocardial potential inverse solutions have been widely used in clinical electrophysiology labs to aid catheter ablation [5], [6]. Currently
such endocardial mapping-based procedures are widely used
for managing atrial arrhythmias, in which the thin wall of atria
Authorized licensed use limited to: University of Minnesota. Downloaded on August 21, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
HE et al.: THREE-DIMENSIONAL CARDIAC ELECTRICAL IMAGING FROM INTRACAVITY RECORDINGS
may explain the current clinical success of the 2-D approach.
The present work is moving one step further from the 2-D endocardial surface mapping to the 3-D cardiac imaging of electrical activity, based on the electrical potential recordings on a
catheter. As the ventricular walls are much thicker than those
of atria, the ability of performing 3-D electrical imaging may
have a significant impact on clinical practice to manage ventricular arrhythmias. The present computer simulation studies have
demonstrated the feasibility of this new approach, and its performance in localizing and imaging site of activation and activation
sequence during single and dual pacing.
The uniqueness of inverse solutions is an important issue
to consider. Without regularization, there would be an infinite
number of inverse solutions of 3-D source distribution for a
given surface potential distribution under the quasi-static condition. The use of regularization enables us to obtain a unique
solution for the inverse problem from a mathematical point of
view. Note that the mathematically unique solution may not
necessarily be the true inverse solution within the heart. Thus,
in this sense any inverse solution shall be considered as an
“estimate” of the true cardiac source distribution, regardless
of the source model being considered. Good estimates can be
obtained by incorporating biophysical or electrophysiological
constraints into the regularization procedure. The idea to use
a heart electrophysiological computer model as regularization
for estimating the 3-D cardiac inverse solutions [9], [10], [19]
is appealing, as it introduces electrophysiological a priori
information into the inverse problem including infracted heart
[20]. On the other hand, if the heart model deviates from the
actual heart then there will be “equivalent noise” being introduced. A previous study dealing with BSPM-based 3-D cardiac
activation imaging suggests that with up to 10% variation in
conduction velocity the approach was still able to image the
overall activation sequence reasonably well [9]. While such
regularization is strong in the sense that it rejects those possible
solution candidates which do not behave reasonably in the
context of cardiac electrophysiology, previous experimental
evidence demonstrated that such 3-D inverse cardiac activation
sequence was well correlated with directly measured 3-D
cardiac activation sequence in an experimental animal model
[17]. An alternative constraint one can introduce is to use the
weighted minimum norm solutions as shown in (1), and then
derive further electrophysiological properties of myocardial
tissues from the estimated current density distribution based
on biophysical constraints. Since such inverse solutions are all
estimates to the true solutions, it is of importance to validate
experimentally the inverse solutions to see if the incorporated
electrophysiological or biophysical constraints do lead to a
reliable estimate of the true solution [17]. While the innovation
of the present work is to image cardiac electrical activity from
intracavity electrical signals, the issue of uniqueness shall remain the same as the BSPM-based 3-D cardiac inverse solution,
as discussed above.
It is of interest to note that, the uniqueness of inverse solutions is not limited to 3-D inverse solutions but also in 2-D inverse solutions, such as the epicardial inverse solutions. For a
2-D epicardial inverse solution, if the number of unknowns is
greater than the number of measurements (currently in the order
of few hundreds), the inverse problem will be underdetermined
and thus there will be an infinite number of inverse solutions. In
1459
order to solve the nonunique underdetermined epicardial inverse
problem, one must also introduce the regularization in order to
obtain a mathematically unique solution. Similar to the 3-D cardiac inverse solutions, there is no guarantee that such regularized epicardial inverse solutions would be the true inverse solution. All we get are “estimates” to the true epicardial inverse solution. When the regularization being introduced is meaningful
in a physiological or biophysical sense, then good estimates are
obtained. Again, such inverse estimates need to be validated experimentally. Regardless of such challenge, progress has been
made in the past decades on the epicardial inverse solutions
[21]–[24].
The ECG inverse solutions from BSPMs have been studied
for decades [19], [21]. In the past few decades, major efforts
have been made to estimate epicardial potentials [21], [22], [25]
and heart surface activation sequence [26]–[28]. Recent efforts
have been made to image the 3-D cardiac electrical activities
from the BSPMs [8]–[10], [17]–[20], [29], [30], and from MCG
[31]. The present study, to our knowledge, represents the first effort in localizing and imaging cardiac electrical activity within
the 3-D volume of the heart from intracavity electrical signals.
Note that while we demonstrated the feasibility of imaging 3-D
cardiac activation from ICPMs by means of the implementation of using heart electrophysiological computer model as regularization, what we proposed is not necessarily limited to this
particular implementation. In other words, other inverse regularization schemes not using a heart electrophysiological computer model (e.g., [29], [32]) may also be used to solve the 3-D
ICPM-based inverse problem. While we only tested the application to imaging of cardiac activation, the proposed 3-D cardiac electrical imaging approach should also be applicable to
imaging of cardiac repolarization. Due to the wide acceptance
of catheter mapping in a clinical setting, we believe the present
proposed approach represents an important advance of cardiac
electrical imaging research. Further experimental validation of
this new approach may establish the approach as a clinically
useful tool for guiding catheter ablation of a variety of cardiac
arrhythmias, aiding rationale determination of optimal locations
of leads for cardiac resynchronization therapy, and other applications to management of cardiovascular disorders.
REFERENCES
[1] R. J. Myerburg, K. M. Dessler, and A. Castellanos, “Sudden cardiac
death: Epidemiology, transient risk, and intervention assessment,” Ann.
Intern. Med., vol. 119, pp. 1187–1197, 1993.
[2] J. P. Dimarco, “Medical progress: Implantable cardioverter defibrillators,” N. Engl. J. Med., vol. 349, pp. 1836–1847, 2003.
[3] W. G. Stevenson and E. Delacretaz, “Radiofrequency catheter ablation
of ventricular tachycardia,” Heart, vol. 84, pp. 553–559, 2000.
[4] L. Gepstein, G. Hayam, and S. A. Ben-Haim, “A novel method for nonfluoroscopic catheter-based electroanatomical mapping of the heart: In
vitro and in vivo accuracy results,” Circulation, vol. 95, pp. 1611–1622,
1997.
[5] C. C. Gornick, S. W. Adler, B. Pederson, J. Hauck, and J. B.
Schweitzer, “Validation of a new noncontact catheter system for electroanatomic mapping of left ventricular endocardium,” Circulation,
vol. 99, pp. 829–835, 1999.
[6] A. Kadish, J. Hauck, B. Pederson, G. Beatty, and C. Gornick, “Mapping
of atrial activation with a noncontact, multielectrode catheter in dogs,”
Circulation, vol. 99, pp. 1906–1913, 1999.
[7] E. O. Velipasaoglu, H. Sun, F. Zhang, K. L. Berrier, and D. S. Khoury,
“Spatial regularization of the electrocardiographic inverse problem and
its application to endocardial mapping,” IEEE Trans. Biomed. Eng.,
vol. 47, no. 3, pp. 327–337, Mar. 2000.
Authorized licensed use limited to: University of Minnesota. Downloaded on August 21, 2009 at 15:14 from IEEE Xplore. Restrictions apply.
1460
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 8, AUGUST 2007
[8] G. Li and B. He, “Localization of the site of origin of cardiac activation by means of a heart-model-based electrocardiographic imaging
approach,” IEEE Trans. Biomed. Eng., vol. 48, no. 6, pp. 660–669, Jun.
2001.
[9] B. He, G. Li, and X. Zhang, “Noninvasive three-dimensional activation
time imaging of ventricular excitation by means of a heart-excitation
model,” Phys. Med. Biol., vol. 47, pp. 4063–4078, 2002.
[10] B. He, G. Li, and X. Zhang, “Noninvasive imaging of cardiac transmembrane potentials within three-dimensional myocardium by means
of a realistic geometry anisotropic heart model,” IEEE Trans. Biomed.
Eng., vol. 50, no. 10, pp. 1190–1202, Oct. 2003.
[11] J. Nenonen, J. Edens, L. J. Leon, and B. M. Horacek, “Computer model
of propagated excitation in the anisotropic human heart: I. implementation and algorithms,” in Computer in Cardiology, 1991, pp. 545–548.
[12] M. Lorange and R. M. Gulrajani, “A computer heart model incorporating anisotropic propagation. I. model construction and simulation of
normal activation,” J. Electrocardiol., vol. 26, pp. 245–261, 1993.
[13] D. Wei, O. Okazaki, K. Harumi, E. Harasawa, and H. Hosaka, “Comparative simulation of excitation and body surface electrocardiogram
with isotropic and anisotropic computer heart models,” IEEE Trans.
Biomed. Eng., vol. 42, no. 4, pp. 343–357, Apr. 1995.
[14] G. Fischer, B. Tilg, R. Modre, G. Huiskamp, J. Fetzer, W. Rucker,
and P. Wach, “A bidomain model based BEM-FEM coupling formulation for anisotropic cardiac tissue,” Ann. Biomed. Eng., vol. 28, pp.
1229–1243, 2000.
[15] Y. Zhang, S. Zhu, and B. He, “A high-order finite element algorithm
for solving the three-dimensional EEG forward problem,” Phys. Med.
Biol., vol. 49, pp. 2975–2987, 2004.
[16] J. Kowalik and M. R. Osborne, Methods for Uniconstrained Optimization Problems. New York: Elsevier, 1968.
[17] X. Zhang, I Ramachandra, Z. Liu, B. Muneer, S. M. Pogwizd, and B.
He, “Noninvasive three-dimensional electrocardiographic imaging of
ventricular activation sequence,” Am. J. Physiol.—Heart Circulatory
Physiol., vol. 289, no. 6, pp. H2724–2732, 2005.
[18] G. Li, X. Zhang, J. Lian, and B. He, “Noninvasive localization of
the site of origin of paced cardiac activation in human by means of
a 3-D heart model,” IEEE. Trans. Biomed. Eng., vol. 50, no. 10, pp.
1117–1120, Oct. 2003.
[19] B. , , B. He, Ed., “Electrocardiographic Tomographic Imaging,” in
Modeling and Imaging of Bioelectric Activity: Principles and Applications. Norwell, MA: Kluwer Academic/Plenum , 2004, pp. 161–182.
[20] G. Li and B. He, “Noninvasive estimation of myocardial infarction
by means of a heart-model-based imaging approach—A simulation
study,” Med. Biol. Eng. Comput., vol. 42, no. 1, pp. 128–136, 2004.
[21] F. Greensite, , B. He, Ed., “Heart surface electrocardiographic inverse
problem,” in Modeling and Imaging of Bioelectric Activity—Principles
and Applications. Norwell, MA: Kluwer Academic/Plenum , 2004.
[22] J. E Burnes, B. Taccadi, and Y. Rudy, “A noninvasive imaging modality
for cardiac arrhythmias,” Circulation, vol. 102, pp. 2152–2158, 2000.
[23] R. C. Barr and M. S. Spach, “Inverse calculation of QRS-T epicardial
potentials from normal and ectopic beats in the dog,” Circ. Res., vol.
42, pp. 661–675, 1978.
[24] R. D. Throne and L. G. Olson, “A generalized eigensystem approach
to the inverse problem of electrocardiography,” IEEE Trans. Biomed.
Eng., vol. 41, no. 6, pp. 592–600, Jun. 1994.
[25] B. He and D. Wu, “A bioelectric inverse imaging technique based
on surface laplacians,” IEEE Trans. Biomed. Eng., vol. 44, no. 7, pp.
529–538, Jul. 1997.
[26] G. Huiskamp and F. Greensite, “A new method for myocardial activation imaging,” IEEE Trans. Biomed. Eng., vol. 44, no. 6, pp. 433–446,
Jun. 1997.
[27] A. J. Pullan, L. K. Cheng, M. P. Nash, C. P. Bradley, and D. J. Paterson,
“Noninvasive electrical imaging of the heart: Theory and model development,” Ann. Biomed. Eng., vol. 29, pp. 817–836, 2001.
[28] R. Modre, B. Tilg, G. Fischer, and P. Wach, “Noninvasive myocardial
activation time imaging: A novel inverse algorithm applied to clinical
ECG mapping data,” IEEE Trans. Biomed. Eng., vol. 49, no. 10, pp.
1153–1161, Oct. 2002.
[29] B. He and D. Wu, “Imaging and visualization of 3D cardiac electric
activity,” IEEE Trans. Inf. Technol. Biomed., vol. 5, no. 3, pp. 181–186,
Sep. 2001.
[30] O. Skipa, M. Nalbach, F. Sachse, C. Werner, and O. Dossel, “Transmembrane potential reconstruction in anisotropic heart model,” Int. J.
Bioelect.-Tromagn., vol. 4, no. 2, pp. 17–18, 2002.
[31] S. Ohyu, Y. Okamoto, and S. Kuriki, “Use of ventricular propagated
excitation model in the magnetocardiographic inverse problem for reconstruction of electrophysiological properties,” IEEE. Trans. Biomed.
Eng., vol. 49, no. 6, pp. 509–519, Jun. 2002.
[32] Z. Liu, C. Liu, and B. He, “Noninvasive reconstruction of three-dimensional ventricular activation sequence from the inverse solution of distributed equivalent current density,” IEEE Trans. Med. Imag., vol. 25,
no. 10, pp. 1307–1318, Oct. 2006.
[33] B. He, “Imaging 3-Dimensional cardiac electrical activity from intracavity potentials,” in Proc. 28th Annu. Int. Conf. IEEE Engineering in
Medicine and Biology Society, 2006, pp. 4519–4519.
Bin He (S’87–M’88–SM’97–F’04) received the
B.S. degree in electrical engineering with the highest
honors from Zhejiang University, Hangzhou, China.
He received the Ph.D. degree in biomedical engineering with the highest honors from the Tokyo
Institute of Technology, Tokyo, Japan, and completed his postdoctoral fellowship in biomedical
engineering at Harvard University—M.I.T.
After working as a Research Scientist at M.I.T.,
he joined the faculty of the University of Illinois
at Chicago (UIC) and became a Professor of Bioengineering, Electrical and Computer Engineering, and Computer Science at
UIC. Since January 2004 he has been a Professor of Biomedical Engineering,
Electrical Engineering, and Neuroscience, and the Director of Biomedical
Functional Imaging and Neuroengineering Laboratory at the University of
Minnesota at Twin Cities. His major research interests include biomedical
functional imaging, neural and cardiac electrical imaging, impedance imaging,
neural interfacing, and bioelectromagnetism. He is the Editor of the books
Neural Engineering (Kluwer , 2005) and Modeling and Imaging of Bioelectric
Activity—Principles and Applications (Kluwer , 2004).
Dr. He is an AIMBE Fellow. He received the NSF CAREER Award, the
American Heart Association Established Investigator Award, the University of
Illinois University Scholar Award, the University of Illinois at Chicago College
of Engineering Faculty Research Award, and the Tejima Prize. He is listed in
Who’s Who in Science and Engineering, Who’s Who in America, and Who’s
Who in the World. He has been active in professional activities in the field
of biomedical engineering. He has served, since 2005, as the Vice President
for Publications of the IEEE Engineering in Medicine and Biology Society,
served as the President of International Society of Bioelectromagnetism
(2002–2005), and has been active in a number of international society and
conference program committees. He serves as an Associate Editor for IEEE
TRANSACTIONS ON BIOMEDICAL ENGINEERING (TBME), IEEE TRANSACTIONS
ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING (TNSRE), IEEE
TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE (TITB), and
the International Journal of Bioelectromagnetism. He is on the editorial board
of Clinical Neurophysiology and the Journal of Neural Engineering.
Chenguang Liu (S’05) received the B.S. and M.S.
degrees in biomedical engineering from Tsinghua
University, Beijing, China, in 2001 and 2003, respectively. Since 2003, he has been working towards
the Ph.D. degree in the Department of Biomedical
Engineering at the University of Minnesota, Minneapolis.
His research interests include biological signal
processing, heart modeling, noninvasive electrocardiographic imaging, and animal and clinical
experimentation.
Yingchun Zhang received the B.S. degree in material processing engineering from Shandong University, Jinan, China, in 1998, the M.S. degree in material processing engineering from Harbin Institute of
Technology, Harbin, China, in 2000, and the Ph.D.
degree in electrical engineering from Zhejiang University, Hangzhou, China, in 2004.
He is currently a Postdoctoral Associate in the Department of Biomedical Engineering at the University of Minnesota, Twin Cities. His research interests
include biomedical modeling and numerical analysis,
biomedical signal processing, and EEG/ECG forward and inverse problem.
Authorized licensed use limited to: University of Minnesota. Downloaded on August 21, 2009 at 15:14 from IEEE Xplore. Restrictions apply.