1388

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 6, JUNE 2010

Analysis of Fractionated Atrial Fibrillation

Electrograms by Wavelet Decomposition
Richard P. M. Houben, Senior Member, IEEE, Natasja M. S. de Groot, and Maurits A. Allessie

AbstractThis study introduces the use of wavelet decomposition of unipolar fibrillation electrograms for the automatic detection of local activation times during complex atrial fibrillation (AF).
The purpose of this study was to evaluate this technique in patients
with structural heart disease and longstanding persistent AF. In 46
patients undergoing cardiac surgery, unipolar fibrillation electrograms were recorded from the right atrium, using a mapping array
of 244 electrodes. In 25 patients with normal sinus rhythm, AF was
induced by rapid pacing, whereas 21 patients were in persistent AF.
In patients with longstanding AF, the atrial electrograms showed
a high degree of fractionation. In each patient, 12 s of AF were
analyzed by wavelet transformation (15 scales). The finest scales
(17) were used to reconstruct a local fibrillation electrogram,
whereas with the coarse scales (915), a far-field signal was generated. With these local and far-field electrograms, the primary fibrillation potentials, due to wave propagation underneath the electrode, could be distinguished from double potentials and multiple
components generated by remote wavefronts. Wavelet transformation resulted in AF histograms with a closely Gaussian distribution
and the automatically generated activation maps showed a good
resemblance with fibrillation maps obtained by laborious manual editing. A special chaining algorithm was developed to detect
multiple components in fractionated electrograms. The degree of
fractionation showed a positive correlation with the complexity of
fibrillation, thus providing an objective quantification of the degree
of electrical dissociation of the atria. Wavelet transformation can
be a useful technique to detect the primary potentials and quantify the degree of fractionation of fibrillation electrograms. This
could enable real-time mapping of complex cases of human AF
and classification of the underlying electropathological substrate.
Index TermsArrhythmias, atrial fibrillation (AF), electrophysiology, mapping, wavelet transform.

I. INTRODUCTION
URRENTLY, atrial fibrillation (AF) affects 2.3 million
patients in the United States and nearly 6 million in Europe. It has been estimated that by the year 2050, approximately
5.6 million Americans will have AF [1]. AF is a disorder of
the elderly, with about 80% occurring in people older than 65
years. The prevalence is about twofold higher in men than in

Manuscript received April 22, 2009; revised September 3, 2009; accepted

November 22, 2009. Date of publication February 5, 2010; date of current
version May 14, 2010. Asterisk indicates corresponding author.
R. P. M. Houben is with the Medtronic Bakken Research Center, Research
and Technology, 6229 GW Maastricht, The Netherlands.
N. M. S. de Groot was with the Department of Physiology, Cardiovascular
Research Institute Maastricht, Maastricht University, 6200 MD Maastricht, The
Netherlands. He is now with Erasmus Medical Center, 3000 DR Rotterdam,
The Netherlands.
M. A. Allessie is with the Department of Physiology, Cardiovascular Research Institute Maastricht, Maastricht University, 6200 MD Maastricht, The
Netherlands (e-mail: m.allessie@fys.unimaas.nl).
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.2009.2037974

women [2][5]. It is characterized by a very rapid and uncoordinated activity of the atria, and a completely irregular ventricular
response. During AF, multiple wavefronts wander through the
atria separated by anatomical and functional barriers. Lines of
block preferentially occur along the border of atrial bundles like
the crista terminalis and the pectinate muscles. In diseased atria
(dilatation, fibrosis), smaller bundles also become dissociated.
Clinically, AF often starts in a paroxysmal form to become more
persistent with time. Although AF is generally not considered
as a life-threatening arrhythmia, it is associated with a considerable increase in mortality [6] and has an important impact on
the quality of life [7].
To diagnose the stage of development of the underlying
pathological substrate of AF, characterization of the underlying spatiotemporal process would be useful. Measurement of
local activation times allows identification of areas with conduction delays and/or conduction block. Previous methods for
automatic identification of local activation times in extracellular
electrograms are mainly based on amplitude, slope, and interval
criteria [8], [9]. Alternatively, we recently developed a template
matching technique for the processing of fibrillation electrograms [10], [11]. As a first step, this technique was validated in
25 patients with normal sinus rhythm, in whom AF was induced
by rapid pacing [12], [13]. 244 Unipolar electrograms were
recorded simultaneously from the free wall of the right atrium
using a slightly concave mapping electrode (diameter 3.6 cm;
interelectrode distance 2.25 mm). A silver plate in the thoracic
cavity was used as an indifferent electrode. All electrograms
were bandpass filtered (0.5500 Hz), sampled with 1 kHz, and
digitized with 12-bit resolution. All electrograms were correlated with a library of 128 mathematically defined extracellular
potentials of different duration and morphology (templates).
This resulted in a correlogram in which positive peaks represented the moments that the fibrillation waves passed under
the electrode. During acutely induced AF, template matching
yielded a very high performance with an overall sensitivity and
positive predictive value of 96.6% 2.5% and 94.3% 5.4%,
respectively. [11] False-positive detection was mainly due to
far-field potentials of remote fibrillation waves and double potentials recorded at the boundaries between waves. Incidental
fractionation of fibrillation electrograms in these patients was
responsible for undersensing of fibrillatory activity. This automatic detection of local fibrillation potentials opened the possibility for real-time imaging in the spatial domain [10]. However,
it is well known that in patients with structural heart disease
and longstanding AF, an increasing number of the fibrillation
electrograms become fractionated [14][16]. Spach and Dolber
have shown that such fractionation is because of asynchronous

0018-9294/$26.00 2010 IEEE

HOUBEN et al.: ANALYSIS OF FRACTIONATED ATRIAL FIBRILLATION ELECTROGRAMS BY WAVELET DECOMPOSITION

activation of neighboring muscle bundles, caused by electrical

uncoupling and a high degree of nonuniform anisotropy of the
atria [17]. Because of the high variation in morphology of fractionated electrograms, template matching becomes impractical
because it would require a very large library of templates.
An alternative technique for the analysis of complex waveforms is to decompose these signals by wavelet transformation. [18], [19]. By correlating fibrillation electrograms with a
single mathematically defined wavelet that is expanded in different time scales, they are decomposed into a subset of signals
that represent the different frequency bands of the fibrillation
electrograms. In our study, these wavelet transforms were used
to improve the specificity of the template matching technique
to detect fibrillation potentials in patients with complex AF. In
addition, wavelet decomposition also provided an estimation of
the degree of fractionation of fibrillation electrograms.
The purpose of our study was twofold: 1) to determine the
performance of template matching in patients with structural
heart disease and longstanding AF and 2) to evaluate the value
of wavelet decomposition for the analysis of fractionated fibrillation electrograms.
II. METHODS
A. Processing of Fractionated Fibrillation Electrograms
In general, the morphology of unipolar fibrillation electrograms can be divided into single, short-double, long-double,
and fractionated potentials [13]. Single potentials are characterized by a single rapid negative deflection preceded by a positive
R-wave and smoothly returning to the baseline (S-wave). Shortdouble potentials are characterized by two negative deflections
separated by less then 15 ms, whereas long-double potentials
consist of two components more than 15 ms apart [13]. Fractionated potentials are composed of three or more dissociated
components, leading to a prolongation of the total activation
complex.
Fig. 1 illustrates the six consecutive steps we used for the
analysis of fractionated fibrillation electrograms. First, the drift
of the baseline was corrected by linear-phase high-pass filtering
(cutoff frequency 1.0 Hz). Reduction of the ventricular components in the unipolar AF electrograms was achieved by subtraction with a QRS template. This template was obtained by
averaging all the 244 unipolar electrograms recorded from the
right atrium. Since during AF, the atrial components are asynchronous and the ventricular QRS complexes have a comparable morphology in the different electrograms, most of the atrial
potentials will be averaged out, resulting in a template of the
spatially averaged ventricular component.
An initial set of local activation times was obtained by template matching, using a relatively low correlation threshold of
0.6 to attain a high sensitivity, but relatively low specificity
(step 1). Then wavelet decomposition was performed using the
first derivative of a Gaussian waveform (step 2). The fine scales
(17) contain the high-frequency components of the electrogram, whereas the coarser scales (815) represent the lower
frequencies. The amplitude of each scale was plotted in a color
scheme, positive correlations ranging from red to yellow and

1389

Fig. 1. Consecutive processing steps for the automatic analysis of unipolar fibrillation electrograms. After QRS subtraction and baseline correction, an initial
set of activation marks was generated by oversensitive template matching (bars).
Then wavelet decomposition was applied, resulting in a scalogram consisting of
15 different scales. Yellow indicates the maxima and black the minima in each
scale. A near-field electrogram was constructed by exponential summation
of the finest seven scales. This near-field electrogram was used to confirm true
fibrillation potentials (red marks), remove false positive detections, and identify double potentials (short blue marks). Multiple components in fractionated
electrograms were then identified by wavelet analysis (short blue marks). As a
last step, fractionated potentials missed by template matching (false negatives)
were added.

negative correlations from red to black (see Fig. 1). These 15

wavelet transforms were used for the reconstruction of a nearfield electrogram by calculating the exponentially weighted
sum of the first seven scales. Similarly, a far-field electrogram was created by summation of the coarse scales 9 to 15
(not shown) (step 3). With these reconstructed near- and farfield electrograms, the true activations were identified (red
marks) and false-positive detections were removed (black dotted marks) (step 4). A special chaining algorithm was then
applied to identify multiple components in fractionated electrograms (short blue marks) (step 5). As a last step, the falsenegative detections of the template matching algorithm were
corrected by accepting one or more of the multiple components
during a long AF cycle as primary activations (red marks).
B. Template Matching
The morphology of single and short-double potentials was
reproduced by constructing a library of mathematically defined
templates [10], [11]. In short, each template was constructed by
piecewise interpolation (1 ms) between four fiducial points indicating the start and peak of the R- and S-waves. First, a set of
27 single potentials was constructed, comprising nine different
RS ratios and three durations (10, 20, and 30 ms). Short-double
potentials were generated by summation of any two of these
27 potentials after a phase shift of 416 ms and varying the
amplitude of one of the potentials. The resulting 7290 permutations were then reduced by a clustering procedure [20] to a final
library of 128 templates. By design, the template library was
limited to single and short-double potentials. It was intended to

1390

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 6, JUNE 2010

represent the physiological variation in morphology of fibrillation potentials in normal atria, due to tissue anisotropy, variation
in conduction velocity, and slight dissociation of neighboring
muscle bundles.
All fibrillation electrograms were matched with all templates
in the library in time steps of 1 ms. This resulted in a correlogram
of all successive maximal normalized correlation coefficients.
The exact moments of activation were determined by scanning
the fibrillation electrograms for the most negative slopes in a
time window of 10 ms around the positive correlation peaks.

TABLE I
CHARACTERISTICS OF THE 15 SCALES OF THE USED GAUSSIAN WAVELET

C. Wavelet Decomposition
Wavelet transformation was introduced 18 years ago by
Daubechies [21], and has been further developed by Mallat [19]
and Strang and Nguygen [18]. Nowadays, this technique is
widely used in various biomedical applications [22][26].
The wavelet we used for decomposition of fibrillation electrograms is defined by
d
2
(t),
where
(t) = et .
(1)
dt
This wavelet was stepwise expanded to 15 different time
scales to calculate the wavelet coefficients


 +
t
1
f (t)
Wf (, s) =

dt or in short
s
|s|
(t) =

d
s (t)
(2)
dt
where Wf are the wavelet coefficients, f the unipolar electrogram, the time shift, and s one of the 15 scales (115).
The

power of the different scales was normalized by the factor
1/ |s|.
The total duration of the 15 expanded wavelets ranged
between 3 and 90 ms. Transformation with these wavelets produces positive peaks at the negative slopes in the fibrillation
electrograms.
A wavelet transform can equally be considered as a bandpass
filter by rewriting formula (2) into
Wf (, s) = f

d
(f s )(t)
(3)
dt
where Wf is the derivative of the electrogram f after smoothing
by a low-pass filter defined by the scaling function s . At the
finest scales, the wavelet transform is equivalent to a bandpass
filter between 250 and 500 Hz (3dB). The intermediate scale
(8) passes frequencies between 17.8 and 61.8 Hz, whereas the
coarsest scale (15) is only sensitive for low frequencies (between
8.3 and 28.8 Hz). The wavelet transform was calculated for
the full length of all recorded fibrillation electrograms. After
calculating the wavelet transform, coefficients were corrected
for filter delays equal to half the duration of the wavelet (see
Table I).
In Fig. 2, four fibrillation maps are shown together with the
fibrillation potentials recorded at the sites indicated by open
circles. In Fig. 2 (bottom), the corresponding scalograms of
the 15 wavelet transforms are plotted. The finest scales contain the higher frequencies of the electrograms and the coarse
Wf (, s) =

Fig. 2. Wavelet decomposition of unipolar fibrillation electrograms. Four

manually reconstructed fibrillation maps, recorded from the epicardial wall
of the right atrium, using a mapping array of 244 electrodes (diameter 3.6 cm,
isochrones at 10 ms). Of each map, one fibrillation potential is shown recorded at
the site indicated by the open circle. At the bottom, the corresponding scalograms
are given. A: Fibrillation potential generated by a broad uniformly propagating fibrillation wave. B: Single fibrillation potential preceded by a ventricular
far-field potential (in volts). C: Double potential generated by two neighboring
fibrillation waves. D: Fractionated electrogram consisting of multiple components generated by multiple wavelets propagating in close proximity to the
recording electrode.

scales the low frequencies. In panel A, a single broad fibrillation wave propagated under the mapping electrode, producing a sharp biphasic electrogram, composed of both near- and
far-field components. Consequently, the signal generated high
amplitudes in all scales of the scalogram. At the fine scales,
the transforms consist of monophasic narrow positive peaks

HOUBEN et al.: ANALYSIS OF FRACTIONATED ATRIAL FIBRILLATION ELECTROGRAMS BY WAVELET DECOMPOSITION

that coincided with the steep negative deflection in the unipolar

electrogram, representing the moments of depolarization of the
tissue under the electrode. In the intermediate and coarse scales,
the wavelet transforms gradually became triphasic in shape, the
positive peaks being preceded and followed by negative potentials of lower amplitude. These broad negative peaks were due to
the electrotonic potentials recorded by an extracellular electrode
when a wavefront, respectively, approaches or propagates away
from the recording site. Panel B shows a map of two colliding
and fusing wavefronts shortly after activation of the ventricles.
In this case, the QRS complex was not subtracted, and depolarization of the ventricles appeared in the atrial electrogram as a
high and broad positive peak (V). Because of its low-frequency
content, this ventricular far-field potential was exclusively expressed in the coarse scales of the scalogram and did not appear
in the finer transforms. In panel C, the atrium was activated
by multiple dissociated depolarization waves. The fibrillation
electrogram, showing a clear double potential, was recorded at
the boundary between two of these waves. The first and steepest negative deflection resulted from the actual passage of one
of these fibrillation waves under the electrode. The second potential was due to electrotonic current generated by the other
wave when passing at close distance. This second far-field potential was mainly expressed in the coarser scales and produced
only low amplitudes in the finer scales. In panel D, atrial activation was highly disorganized and the atrium was activated
by many narrow wavefronts propagating in different directions.
The multiple components of the fractionated fibrillation electrogram were predominantly expressed in the intermediate scales
of the wavelet transforms.

D. Construction of Near- and Far-Field Electrograms

Unipolar fibrillation electrograms are composed of local potentials generated by depolarization waves passing under the
electrode and far-field potentials of other more remote fibrillation waves. The factors that determine the shape of the extracellular potential include the spatial extent and propagation
velocity of the wavefronts, and the distance to the recording
electrode. According to the solid angle theorem, the amplitude
of an approaching wavefront is proportional to the solid angle
seen by the extracellular electrode [27]. For distal wavefronts,
the solid angle first increases at a moderate rate, resulting in a
low-frequency foot of the R-wave. The nearer the wavefront
gets to the recording electrode, the faster the solid angle will
grow and the higher the frequency content of the R-wave will
be. At the time the depolarization wave is passing under the electrode, the polarity of the bipolar potential is suddenly switched,
causing the steep negative deflection (high frequencies) in the
extracellular electrogram. As the negative tail of the wavefront
is retreating from the electrode, the solid angle progressively
shrinks again, leading to an exponential decrease in frequency of
the S-wave. Thus, the local components in the electrogram will
be mainly expressed in the fine scales of the wavelet transform,
whereas the far-field signals are predominantly represented in
the coarse scales.

1391

Fig. 3. Creation of near- and far-field electrograms. Near- and far-field electrograms are calculated by weighed summation (weight factor between 0.11
and 1.0) of the seven finest (17) and seven coarsest scales (915). The fibrillation electrogram contains a single potential (A), a short double potential (B),
and a fractionated potential (C). In the reconstructed near-field electrogram,
the rapid negative deflections of the fibrillation electrogram are enhanced. The
reconstructed far-field electrogram was typically triphasic in shape, the time
between the peak and troughs being determined by the conduction velocity of
the waves. Broad and rapidly propagating wavefronts produced large far-field
electrograms. The far-field components of the fractionated potential (C) were of
low amplitude, indicating that they were generated by smaller fibrillation waves
propagating at close distance from the recording electrode.

To help distinguish local from far-field potentials, a near

and far-field electrogram was reconstructed from the 15
wavelet transforms (see Fig. 3). Since the expression of the
local and distant wavefronts in the different scales is gradual,
the wavelet transforms were weighed exponentially as follows:

Wf (t, s)Enear (s)
rnear (t) = s 
s Enear (s)

Wf (t, s)Efar (s)
(4)
rfar (t) = s 
s Efar (s)
where rnear and rfar indicate the amplitude of the near- and farfield electrograms, s the scale of the wavelets, Wf the amplitude
of the wavelet coefficients, and E the exponential weight function. These functions exponentially reduce the weight of scales
17 for the reconstruction of a near-field electrogram, and progressively increase the weight of the coarse scales (915) for the
far-field electrogram. They result in a frequency band of 34.7
500 Hz (3dB) for the near-field electrogram and 10.142.4 Hz
for the far-field electrogram.
E. Removal of False Positives
The far-field potentials generated by remote fibrillation waves
were removed by an algorithm using the relative amplitudes of
the near- and far-field electrograms. Potentials were accepted as
primary potentials, expressing the actual passage of a fibrillation wave under the electrode, when they satisfied the following
condition:
rnear > (rnear m in + (rnear m ax rnear m in )(1 er f a r / ))
(5)

1392

Fig. 4. AFCL histograms of the right atrial wall in three patients with an
increasing degree of disorganization of AF. The black histograms on the left
were generated by template matching alone. On the right, the AFCL histograms
are shown after wavelet filtering. The red histograms represent the distribution
of the true fibrillation intervals, obtained after removal of the false-positive
detections resulting from template matching (gray subpopulation). The subpopulation of long AF cycles (black) was due to undersensing of fractionated
fibrillation potentials.

where rnear and rfar indicate the amplitude of the near- and farfield electrograms as defined in (4), rnear m in the minimally required near-field amplitude, and rnear m ax the value above which
fibrillation potentials are accepted, regardless of the amplitude
of their far-field electrogram. The parameter determines the
acceptance rate of low amplitudes in the near-field electrogram, depending on the corresponding amplitude in the far-field
electrogram.
The proportion of false intervals in the AF cycle length
(AFCL) histograms obtained by template matching was estimated as the difference with the best Gaussian fit of the histograms. The rational for Gaussianity of the AFCL distribution
is given by our observation that after accurate editing of the fibrillation maps, the AFCL histogram showed a close to normal
distribution. The Gaussian fit was made from the right side of the
AFCL histogram that did not contain false-positive values. The
difference between the left side of the AFCL histogram and this
Gaussian fit was used to estimate the number of false-positive
detections by template matching.
For each patient, the wavelet filtering parameters and
rnear m ax of (5) were determined as follows. Starting at a low
rnear m ax , the parameter was stepwise increased until the number of removed activations was larger than the estimated number
of false short intervals. As long as this criterion was not satisfied, the procedure was repeated at progressively higher values
of rnear m ax . In Fig. 4, the result of wavelet filtering is shown for
three patients. The black histograms on the left-hand side are the
result of template matching alone, whereas the red histograms
on the right-hand side show the distribution of AF cycles after
removal of false-positive intervals (gray) by wavelet filtering.
The number of activations removed by wavelet filtering could
be set as a percentage of the estimated amount of false-positive

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 6, JUNE 2010

Fig. 5. Effects of different degrees of wavelet filtering. AFCL histograms of

patient 2 are shown after wavelet filtering with 25%175% of the nominal setting. The gray part of the histograms represents the AF cycles that disappear
by progressive filtering of the false-positive fibrillation potentials detected by
template matching. The red histograms represent the true spatiotemporal distribution of AF cycle lengths in the right atrial wall. At nominal filter settings
(100%), a nearly Gaussian distribution of the AF intervals was obtained. Filtering with higher than nominal settings resulted in progressive narrowing of
the AFCL histogram (increase in specificity). This was associated with only
a slight reduction in the amplitude of the AFCL histogram (sensitivity). The
black portion of the histograms represents long AF cycles due to undersensing
of fractionated electrograms.

intervals. At a nominal setting of 100%, the number of removed

detections was equal to the estimated amount of false short
intervals. At higher filter settings, a higher percentage, and at
lower settings, a lower percentage of the false-positive detections by template matching were removed (see Fig. 5). In this
way, the sensitivity and specificity of wavelet filtering could be
selected.
After wavelet filtering of the false-positive fibrillation potentials, the electrograms still contained some double potentials
recorded at the boundaries between fibrillation waves. The amplitude of the double potentials in the surrogate near-field electrogram was used to decide which of the two was generated
by a passing fibrillation wave and which was regarded as an
electrotonic potential. When the interval of a double potential is
longer, it gets more likely that they are caused by two successive depolarizations passing under the electrode. Therefore, the
criteria to accept both potentials as primary activations were
made proportional to the interval between the double potentials.
F. Multiple Components
As the development of the pathological substrate of AF progresses, fibrillation electrograms become more and more fractionated due to electrical uncoupling of adjacent muscle bundles.
Since fractionated potentials are generally of low amplitude and
consist of multiple components, they were often not detected by
template matching. However, fractionated electrograms did produce distinctive peaks in the fine scales of the wavelet transform.

HOUBEN et al.: ANALYSIS OF FRACTIONATED ATRIAL FIBRILLATION ELECTROGRAMS BY WAVELET DECOMPOSITION

They can be distinguished from high-frequency noise because

they penetrated deeper into the intermediate scales of the scalogram. On the basis of this property, a chaining algorithm was
developed that links the positive peaks in the successive wavelet
scales. First, the finest scale was scanned for the presence of
positive peaks. When a positive peak was found, the next scale
was searched for the presence of an associated peak (within a
window of 8 ms). In case a positive peak was found, it was added
to the chain and used as the center of the scanning window for
the next scale. This was repeated until no positive peak was
found or the last scale was reached. The amount of high and
mid-frequencies of the chain was calculated from the amplitude
of the positive peaks in scale 17 with the weighing function
given in (4). In case the amount of high and mid-frequencies of
the chain was higher then a certain threshold value, the chain
was accepted as a multiple component. The moment of the peak
in scale 1 was taken as the time of the multiple components.
In case the interval between two successive chains was less
than 10 ms, only the component with the highest frequency
content was taken. By varying the threshold for multiple components, the sensitivity for quantification of fractionation can be
selected. In our analysis, we used a fixed threshold given by the
fifth percentile of the high and mid-frequencies of all primary
potentials.
G. Correction of False Negatives
As a last step of our algorithm, the activations that had been
missed by template matching were corrected. All AF intervals
longer than twice the median AF interval were checked. If one
of the multiple components exceeded a preset value (tenth percentile of the amplitude of all primary potentials in the reconstructed near-field electrograms), it was accepted as a primary
fibrillation potential.
III. RESULTS
A. Automatic Detection of Fibrillation Potentials
The algorithm for detection of fibrillation potentials was evaluated in 46 patients. About half of them (21) were in persistent
AF for more than one year, and had dilated left atria due to
valvular heart disease. The other 25 patients were in normal sinus rhythm, and AF was induced by rapid atrial pacing. During
open chest surgery, 244 unipolar fibrillation electrograms were
recorded simultaneously from the right atrium. In patients with
longstanding AF, the fibrillation maps were far more disorganized and showed multiple narrow fibrillation waves separated
by lines of conduction block. In Fig. 4, the AFCL histograms
of three representative patients are plotted. In patient 1 (organized AF), AF was induced acutely, whereas in patients 2 and
3 (longstanding AF), AF was, respectively, disorganized and
highly disorganized. The black histograms on the left-hand side
are the result of template matching, whereas the red histograms
on the right-hand side give the estimated distribution of AF cycles after wavelet decomposition. During acutely induced AF,
template matching provided satisfactory results and the population of fibrillation intervals showed a more or less Gaussian

1393

distribution. Wavelet filtering only resulted in the rejection of

a relatively small number of activations (gray subpopulation).
The remaining small subpopulation of long AF intervals (black)
was mainly due to low-amplitude signals. In patients with longstanding AF, template matching was less satisfactory. Because
AF had become disorganized, the multiple fibrillation waves
produced more far-field potentials, thereby resulting in a higher
proportion of false-positive detections. In patient 3, exhibiting
the highest degree of atrial dissociation, this even led to a false
bimodal distribution of AFCLs (lower left panel). The left peak
of the histogram was completely due to oversensing of farfield potentials and did not represent a subpopulation of true
short fibrillation intervals. On the basis of our mapping data,
we defined true fibrillation intervals as intervals that were due
to successive wave propagation under the electrode and not to
an electrotonic source of a more distant depolarization front.
As shown on the right-hand side, wavelet filtering effectively
removed a large part of the population of false short intervals
(gray). The resulting red histograms showed a closer Gaussian
distribution and a better approximation of the population of
true AF cycle lengths.
Fig. 5 illustrates how the balance between false-positive and
false-negative detections (sensitivity/specificity) can be varied
by changing the degree of wavelet filtering (patient 2). From the
top-left to the bottom-right panels, the amount of filtering was
increased in steps of 25%. This resulted in the removal of more
and more short intervals (gray subpopulation) and narrowing of
the distribution of estimated AFCLs (red). At a nominal filter
setting of 100% (center), the estimated AF cycles attained a
Gaussian distribution closely. At higher filter settings (bottom
panels), the specificity of the detection of fibrillation potentials
was further increased. As expected, this was associated with a
reduction in sensitivity, illustrated by the slight reduction in amplitude and the progressive narrowing of the red histograms. The
narrowing of the population of estimated AF cycles thus should
not be interpreted as a lower degree of AFCL dispersion, but is
the result of the higher specificity of the detection algorithm.
B. Automatic Mapping of Longstanding Persistent AF
The combination of template matching and wavelet filtering
was able to generate high-quality maps of AF (see Fig. 6). The
maps shown at the top were reconstructed by laborious manual annotation of the fibrillation potentials by an investigator
experienced in mapping of AF (MA), being blinded for the
automatically generated maps. In case of double potentials or
fractionated electrograms, the potential with the steepest slope
and/or highest amplitude was taken as the moment of local activation. Each color represents a time period of 10 ms in the
sequence, as given by the color scheme at the bottom. During
acute AF (patient 1), the fibrillation waves were broad, and in
the map shown, a single fibrillation wave propagated from the
bottom-right to the top-left parts of the map (70 ms from red
to blue). In patients 2 and 3, the pattern of electrical activation
was dissociated and multiple wavefronts propagated through the
atrial wall. At the bottom, the corresponding maps are shown
generated by automatic detection of the fibrillation potentials

1394

Fig. 6. (Top) Manually and (bottom) automatically reconstructed fibrillation

maps in three patients with AF. 244 unipolar fibrillation electrograms were
recorded simultaneously from the free wall of the right atrium (interelectrode
distance 2.25 mm). Local activation times are indicated by colors (classes 10 ms)
as indicated by the color scheme at the bottom. During acutely induced AF, manually reconstructed and automatically generated activation maps were highly
similar (patient 1). In patients with longstanding AF (patients 2 and 3), fibrillation maps were far more complex and showed multiple wavelets propagating
simultaneously in different directions of the atrial wall. However, even during
these cases of highly disorganized AF, the automatically generated maps showed
a remarkable resemblance with the laboriously edited fibrillation maps.

(nominal filter settings). In the patient with acute AF (patient

1), the manual and automatic maps were almost identical (FP =
5.6%, FN = 7.6%). But, also in the more complex cases of
AF (patient 2 and 3), the fibrillation maps showed a close resemblance [false positive (FP) = 9.1% and 7.6%, false negative
(FN) = 14.8% and 9.6%]. Most of the colors in the manually
edited and automatically generated maps were identical. Movies
of manually and automatically generated fibrillation maps can
be viewed on our website www.afdiagnostics.eu.
C. Analysis of Fractionated Potentials
Fig. 7 shows some examples of the analysis of fractionated
electrograms. Below each unipolar electrogram, the primary
activations, resulting from template matching and wavelet filtering, are shown as red marks. The fibrillation electrogram at
the top is of moderate complexity and contains some ventricular far-field potentials that were completely ignored by the
detection algorithm. The chaining algorithm of the wavelet
transforms detected some multiple components, as indicated
by the short blue bars. The second tracing shows a mixture
of high-amplitude single potentials, one double potential, and
a long fractionated complex in the middle (encircled). During
this short period of continuous electrical activity, the detection
algorithm assigned three potentials as primary activations and
five as multiple components. The third electrogram contains
three fractionated complexes of varying degree and duration.
Of each fractionated complex, only one potential was assigned
as a primary activation. The small potential in the long interval
between the third and fourth primary potentials did not meet the
criteria for a primary activation. At the bottom, a low-amplitude

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 6, JUNE 2010

Fig. 7. Four examples of fractionated unipolar fibrillation electrograms. Below each electrogram, the output of the automatic detection algorithm is plotted.
Red bars indicate primary fibrillation potentials. Double potentials and multiple components are indicated by short blue bars. Fractionated portions of the
electrograms are encircled. V: far-field potentials of the ventricles.

electrogram is shown, either due to poor contact or the presence

of scar tissue under the electrode. The two ventricular complexes (V) were ignored and only two atrial components were
found by the detection algorithm, one of them being accepted
as a primary activation.
In Fig. 8, the degree of fractionation is quantified for the
three representative cases. On the left-hand side, the incidence
of double and fractionated potentials is plotted. In the patient
with acute AF (top), 14% of the fibrillation potentials were
fractionated, consisting almost exclusively of double potentials
(12.6%). In the patients with longstanding AF, the degree of fractionation was considerably higher and comprised, respectively,
30% and 42% of all fibrillation potentials. Fractionated potentials now consisted of multiple components (triplets or higher)
more frequently. In Fig. 8 (right-hand side), the duration of all
fractionated potentials is plotted. The red histograms show the
distribution of the primary AF intervals, whereas the delays
between double potentials are plotted in blue. The duration of
the more fractionated complexes, consisting of three or more
components, are plotted in light blue and purple. The interval
between the double potentials varied widely, the longest double
potentials being as long as the shortest primary AF cycles. The
total duration complexes with higher degree of fractionation
ranged between 30 and >300 ms.
D. Validation of Wavelet Decomposition of Fibrillation
Electrograms in 46 Patients
We validated the value of wavelet decomposition for the analysis of complex fibrillation electrograms in 46 patients. A total of 21 patients were in persistent AF for more than a year,
whereas in 25 patients, AF was induced by rapid pacing. The
results are shown in Fig. 9. In Fig. 9 (top), the added value of

HOUBEN et al.: ANALYSIS OF FRACTIONATED ATRIAL FIBRILLATION ELECTROGRAMS BY WAVELET DECOMPOSITION

Fig. 8. Quantification of nature and degree of fractionation of fibrillation

potentials in three patients. In patient 1, AF was induced acutely by rapid
pacing. Patients 2 and 3 were in persistent AF for more than one year. (Left)
Incidence of primary potentials (red), double potentials (dark blue), and higher
order of fractionation (light blue and purple). (Right) Distribution of AF cycle
lengths (red), intervals between double potentials (dark blue), and total duration
of fractionated complexes (light blue: triplets; purple: quads).

wavelet decomposition for the determination of primary fibrillation potentials is evaluated by the degree of Gaussianity of
the obtained AFCL histograms. Template matching alone gave
highly variable results. Especially in the more complex cases
of AF, this technique failed to produce a normal distribution
of fibrillation intervals (Gaussianity <70%). Wavelet decomposition improved the detection of primary fibrillation potentials
considerably. On average, wavelet filtering increased the Gaussianity of the obtained AFCL histograms from 66.5% 19.3%
to 89.1% 10.4% (p < 0.001).
We compared the sensitivity and specificity of template
matching alone, and the combination of template matching and
wavelet decomposition in the whole population of AF patients.
A dataset of carefully edited activation times was used as a
gold standard. As expected, in patients with acutely induced
AF in whom AF was still relatively organized, the added value of
wavelet analysis was limited, with a sensitivity of 96.8% 2.3%
and 90.8% 7.3%, respectively, and a specificity of 90.1%
4.3% and 97.8% 2.7%, respectively. In contrast, in patients
with longstanding persistent AF, the specificity of detection of
fibrillation potentials was largely enhanced. In half of these
patients, showing the highest complexity of AF, specificity improved from 54.6% 8.1% to 78.0% 5.2% (p < 0.001).
In the other half, exhibiting still relatively organized AF, the
specificity increased from 77.6% 5.1% to 85.3% 7.0%
(p < 0.001).
In Fig. 9 (bottom), the number of multiple components detected by wavelet transformation is plotted against the percentage of intra-atrial conduction block (interelectrode conduction
velocity of <19 cm/s). In patients with longstanding AF (closed

1395

Fig. 9. Validation of the automatic analysis of unipolar fibrillation potentials

in 46 patients (21 with longstanding and 25 with acute AF). (Top) Degree of
Gaussianity of the AFCL histograms during acute (open symbols) and persistent
AF (solid symbols) resulting from template matching alone and after additional
wavelet decomposition. The use of wavelet transformation improved the Gaussianity of the detected AF cycles considerably from 66.5% 19.3% to 89.1%
10.4% (p < 0.001). (Bottom) Percentage of fractionated potentials detected by
wavelet transformation during acute AF (open symbols) and longstanding persistent AF (solid symbols). The average degree of fractionation during acute AF
was 13.7% 10.1% compared to 37.1% 13.3% during persistent AF (p <
0.001). On the abscissa, the amount of intraatrial conduction block is plotted,
measured as the percentage of interelectrode conduction velocities of <19 cm/s.
There was a positive correlation between the amount of intraatrial conduction
block during AF and the amount of fractionation.

symbols), the amount of fractionation was about two to three

times higher than during acutely induced AF (37.1% 13.3%
versus 13.7% 10.1%; p < 0.001). As expected, the degree of
fractionation showed a positive correlation with the amount of
intraatrial conduction block (slope: 1.61; r: 0.64; p < 0.001).
IV. DISCUSSION
A. Wavelet Analysis of AF Electrograms
We used the continuous wavelet transform and the first derivative of a Gaussian wavelet to analyze unipolar fibrillation electrograms. Although a more efficient and nonredundant analysis
can be obtained by discrete wavelet transformation, this technique is more sensitive to small shifts in the input signal that may
cause large changes in the calculated wavelet coefficients [28].
The first derivative of a Gaussian waveform was chosen because it closely resembles the basic morphology of unipolar
extracellular potentials. The range of the used 15 scales covers the whole frequency content of fibrillation electrograms.
Unipolar electrograms are the sum of all extracellular potentials
generated by membrane currents of the myocardium around the
recording electrode. Although the contribution of these currents

1396

rapidly decreases with distance, large remote wavefronts still

contribute considerably to unipolar fibrillation electrograms.
Direct passage of a depolarization wave under the electrode
generates the steep (intrinsic) negative deflection in the electrogram, whereas the slower R- and S-waves are due to the
electrotonic potentials associated with a propagating wavefront.
A unipolar fibrillation electrogram thus can be considered as the
sum of all current sources subjected to a spatial low-pass filter,
i.e., the closer to the recording site, the higher the amplitude
and frequency contribution of the current sources [29]. The fine
scales of the wavelet transform cover the steep deflections of
local current sources, whereas the coarse scales represent the
slower electrotonic potentials of more distal sources. Depending on the conduction velocity of the wavefront, the duration of
the intrinsic negative deflection may vary between 1 and 8 ms.
This predominantly produces positive peaks in scales 7 of the
wavelet transforms (see Table I). Since the weight function, used
to reconstruct the near-field electrogram, exponentially reduces
the contribution of scales 17, it favors detection of the more
rapidly propagating wavefronts (>30 cm/s). Thus, the reconstructed near-field electrogram specifically contains information
of fibrillation waves propagating within the normal physiological range (30100 cm/s). The far-field electrogram reconstructed
from the coarse scales (915), primarily provides information
about the width of the fibrillation waves. Botteron and Smith
have shown that the activation space constant during human
AF varies between 1.4 and >4.0 cm [30]. The wavelet transform decomposes the fibrillation potentials into fine and coarse
scales characterized by progressively lower frequency bands.
Surrogate near- and far-field electrograms were constructed by
exponential weighing of the first 7 and the last 7 scales. The
frequency bands used were based on the biophysical properties
of extracellular cardiac electrograms and their variation in shape
and frequency content during AF. The weighing function was
determined empirically, and the parameter settings and functions
involved might need further refinement when they are used on
a wider scale. This could include different settings in different
groups of patients or different types of recording electrodes.
In general, fibrillation electrograms are generated by three
types of waves. 1) Broad wavefronts that propagate under the
electrode. These waves produce extracellular potentials that contain both low frequencies (far-field component) as well as high
frequencies caused by the local passage of the depolarization
wave. 2) Large wavefronts that do not propagate under the electrode, but are passing at some distance. In this case, the solid
angle between the wave and the electrode will not increase
rapidly, and the electrogram only consists of low and intermediate frequencies. 3) Narrow wavefronts in electrically dissociated
muscle bundles. Because of the small size of these waves they
will only be picked up in close vicinity of the electrodes. Farfield components will be small or absent, and consequently, the
electrogram predominantly consists of high frequencies, producing spiky potentials of short duration. When atrial bundles
in the neighborhood are activated without actually propagating
under the electrode, the electrogram will mainly consist of intermediate frequencies. Narrow wavefronts will be predominantly
expressed in the fine scales of the wavelet transform, whereas

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 6, JUNE 2010

remote fibrillation waves will be primarily represented in the

coarse scales. Broad fibrillation waves propagating under the
electrodes will generate high amplitudes in the fine and coarse
wavelet transforms. Thus, wavelet transformation of fractionated unipolar fibrillation electrograms does provide information
about the nature and spatial distribution of the fibrillation waves.
B. Other AF Signal Processing Techniques
In other studies, fast Fourier analysis has been used to estimate the dominant frequency and irregularity of fibrillation
electrograms [31][35]. The obtained frequency spectra of the
fibrillation electrograms are determined by the AF intervals and
the frequency content of the fibrillation potentials. The peak of
the frequency spectra represents the average local AF intervals,
whereas the width is used as an index for the temporal irregularity in cycle length. However, broad frequency spectra can also be
the result of fractionated electrograms. In these cases, the peak
of the spectra is less prominent and might not accurately reflect
the local AF frequencies. Recently, the degree of organization
of AF was quantified by the spatiotemporal distribution of the
fibrillatory wave complexity of bipolar electrograms [36]. With
this method, indexes for the regularity and electrical coupling
were derived, enabling the evaluation of the fibrillatory process
in the temporal and spatial domain. The measurement of the
degree of organization was on the basis of the assumption that
organized activation patterns produce repetitive waveforms of
similar morphology.
C. Clinical Applications
There are two important potential applications of automatic
fibrillation electrogram analysis.
1) The automatic detection of primary fibrillation potentials,
representing local activation of the myocardium, will allow real-time imaging of the fibrillation process. Movies
of AF will not only reveal the number of fibrillation waves
in different parts of the atria, but they will also exhibit spatiotemporal differences in the amount of intraatrial conduction block. This may be used to quantify the amount
of longitudinal dissociation during AF, a crucial element
of the underlying electropathological substrate. Diagnosis
of the electropathological substrate of AF is important for
choosing the optimal treatment strategy for AF of different
etiologies and duration.
2) Ablation of areas with complex fractionated atrial electrograms (CFAEs) has been recently proposed as a treatment
of patients with longstanding persistent AF [16], [37].
At present, the characterization of CFAEs is based on a
variety of parameters, including the percentage of continuous electrical activity, the maximal bipolar voltage,
a fractionation index (number of deflections >0.05 mV),
the temporal activation gradient (time difference 70 ms
between proximal and distal bipoles), and the dominant
frequency [37], [38]. Various studies investigated which
characteristics of CFAEs are most predictive for slowing and termination of AF [39], [40]. Recently, Takahashi
et al. concluded that only the percentage of continuous

HOUBEN et al.: ANALYSIS OF FRACTIONATED ATRIAL FIBRILLATION ELECTROGRAMS BY WAVELET DECOMPOSITION

activity and the presence of a temporal gradient should be

used as a guide for ablation [38].
In clinical practice, intracardiac bipolar recordings are usually preferred above unipolar electrograms because they are
less sensitive to interference and contain smaller far-field potentials. However, the disadvantage of bipolar electrograms is
that they are the sum of two unipolar electrograms recorded at
different sites. Once combined, the individual signals can no
longer be subtracted from the bipolar electrogram. The morphology of bipolar electrograms is largely determined by the
phase differences and direction of wave propagation under the
two poles. This explains why bipolar electrograms of AF show
a much higher spatiotemporal variation in morphology and a
far higher degree of fractionation than unipolar electrograms.
This limits the translation of bipolar fractionated electrograms
into the spatial domain and the discrimination between active and passive CFAEs [41]. The main reason that unipolar
electrograms are rarely used clinically is that they lack commonmode rejection and contain high-amplitude far-field potentials.
Decomposition of unipolar fibrillation electrograms into their
near- and far-field potentials by wavelet decomposition could
facilitate the use of unipolar electrograms for the analysis of
fractionation during AF.

[7]
[8]

[9]

[10]
[11]
[12]

[13]

[14]
[15]

V. CONCLUSION
The automatic analysis of fibrillation electrograms provides
an objective evaluation of fibrillation electrograms and visualization of the complex spatiotemporal processes during AF.
Since the calculations involved in the various algorithms are
not very complex, both template matching and wavelet transformation can be easily implemented in field programmable gate
arrays. This will provide real-time movies of the fibrillatory conduction patterns during AF, and an unbiased quantification of
the nature and degree of fractionated fibrillation electrograms.
This may open the way for a systematic diagnosis of the electropathological substrate of persistent AF in humans.

[16]

[17]

[18]
[19]

REFERENCES

[20]

[1] S. L. Kopecky, B. J. Gersh, M. D. McGoon, J. P. Whisnant, D. R. Holmes,

Jr., D. M. Ilstrup, and R. L. Frye, The natural history of lone atrial
fibrillation. A population-based study over three decades, N. Engl.
J. Med., vol. 317, no. 11, pp. 669674, Sep. 1987.
[2] C. D. Furberg, B. M. Psaty, T. A. Manolio, J. M. Gardin, V. E. Smith, and
P. M. Rautaharju, Prevalence of atrial fibrillation in elderly subjects (the
cardiovascular health study), Amer. J. Cardiol., vol. 74, no. 3, pp. 236
241, Aug. 1994.
[3] W. M. Feinberg, J. L. Blackshear, A. Laupacis, R. Kronmal, and R. G. Hart,
Prevalence, age distribution, and gender of patients with atrial fibrillation.
Analysis and implications, Arch. Intern. Med., vol. 155, no. 5, pp. 469
473, Mar. 1995.
[4] A. S. Go, E. M. Hylek, K. A. Phillips, Y. Chang, L. E. Henault, J. V. Selby,
and D. E. Singer, Prevalence of diagnosed atrial fibrillation in adults:
National implications for rhythm management and stroke prevention: The
AnTicoagulation and risk factors in atrial fibrillation (ATRIA) study,
J. Amer. Med. Assoc., vol. 285, no. 18, pp. 23702375, May 2001.
[5] J. Friberg, H. Scharling, N. Gadsboll, and G. B. Jensen, Sex-specific
increase in the prevalence of atrial fibrillation (The Copenhagen City
Heart Study), Amer. J. Cardiol., vol. 92, no. 12, pp. 14191423, Dec.
2003.
[6] E. J. Benjamin, P. A. Wolf, R. B. DAgostino, H. Silbershatz, W. B. Kannel, and D. Levy, Impact of atrial fibrillation on the risk of death: The

[21]
[22]

[23]
[24]

[25]

[26]
[27]

1397

framingham heart study, Circulation, vol. 98, no. 10, pp. 946952, Sep.
1998.
J. E. Ware, K. K. Snow, M. Kosinski, and B. Gandek, SF-36 Health Survey
Manual and Inter Pretation Guide. Boston, MA: New England Medical
Center, The Health Institute, 1993.
D. Scherr, D. Dalal, A. Cheema, A. Cheng, C. A. Henrikson, D. Spragg,
J. E. Marine, R. D. Berger, H. Calkins, and J. Dong, Automated detection
and characterization of complex fractionated atrial electrograms in human
left atrium during atrial fibrillation, Heart Rhythm, vol. 4, no. 8, pp. 1013
1020, Aug. 2007.
M. Porter, W. Spear, J. G. Akar, R. Helms, N. Brysiewicz, P. Santucci, and
D. J. Wilber, Prospective study of atrial fibrillation termination during
ablation guided by automated detection of fractionated electrograms,
J. Cardiovasc.. Electrophysiol., vol. 19, no. 6, pp. 613620, Jun. 2008.
R. P. Houben and M. A. Allessie, Processing of intracardiac electrograms
in atrial fibrillation. Diagnosis of electropathological substrate of AF,
IEEE Eng. Med. Biol. Mag., vol. 25, no. 6, pp. 4051, Nov. 2006.
R. P. Houben, N. M. de groot, F. W. Lindemans, and M. A. Allessie,
Automatic mapping of human atrial fibrillation by template matching,
Heart Rhythm., vol. 3, no. 10, pp. 12211228, Oct. 2006.
K. T. Konings, C. J. Kirchhof, J. R. Smeets, H. J. Wellens, O. C. Penn,
and M. A. Allessie, High-density mapping of electrically induced atrial
fibrillation in humans, Circulation, vol. 89, no. 4, pp. 16651680, Apr.
1994.
K. T. Konings, J. L. Smeets, O. C. Penn, H. J. Wellens, and M. A. Allessie,
Configuration of unipolar atrial electrograms during electrically induced
atrial fibrillation in humans, Circulation, vol. 95, no. 5, pp. 12311241,
Mar. 1997.
K. Nademanee, M. Schwab, J. Porath, and A. Abbo, How to perform
electrogram-guided atrial fibrillation ablation, Heart Rhythm., vol. 3,
no. 8, pp. 981984, Aug. 2006.
J. Lin, B. J. Scherlag, J. Zhou, Z. Lu, E. Patterson, W. M. Jackman,
R. Lazzara, and S. S. Po, Autonomic mechanism to explain complex
fractionated atrial electrograms (CFAE), J. Cardiovasc. Electrophysiol.,
vol. 18, no. 11, pp. 11971205, Nov. 2007.
H. Oral, A. Chugh, E. Good, A. Wimmer, S. Dey, N. Gadeela, S. Sankaran,
T. Crawford, J. F. Sarrazin, M. Kuhne, N. Chalfoun, D. Wells, M. Frederick, J. Fortino, S. Loucif-Moore, K. Jongnarangsin, F. Pelosi, Jr., F. Bogun,
and F. Morady, Radiofrequency catheter ablation of chronic atrial fibrillation guided by complex electrograms, Circulation, vol. 115, no. 20,
pp. 26062612, May 2007.
M. S. Spach and P. C. Dolber, Relating extracellular potentials and their
derivatives to anisotropic propagation at a microscopic level in human
cardiac muscle. Evidence for electrical uncoupling of side-to-side fiber
connections with increasing age, Circ. Res., vol. 58, no. 3, pp. 356371,
Mar. 1986.
G. Strang and T. Nguygen, Wavelets and Filterbanks. Cambridge, MA:
Wellesley-Cambridge, 1996, pp. 1490.
S. Mallat, A Wavelet Tour of Signal Processing, 2nd ed. San Diego, CA:
Academic, 1998, pp. 1577.
J. H. Ward, Hierarchical grouping to optimize an objective function,
J. Amer. Statist. Assoc., vol. 58, pp. 236244, 1963.
I. Daubechies, Ten Lectures on Wavelets. Philadelphia, PA: SIAM, 1992.
J. G. Akar, T. H. Everett, L. C. Kok, J. R. Moorman, and D. E. Haines, Effect of electrical and structural remodeling on spatiotemporal organization
in acute and persistent atrial fibrillation, J. Cardiovasc. Electrophysiol.,
vol. 13, no. 10, pp. 10271034, Oct. 2002.
J. Liu and Q. Zhang, [Detection of ECG characteristic using wavelet
transform], Zhongguo Yi. Liao. Qi Xie. Za Zhi., vol. 23, no. 6, pp. 324
327, Nov. 1999.
H. Li, D. Hao, and Y. Duan, [An approach to achieve classification of
QRS waves with wavelet multiresolution analysis and fractal dimension],
Sheng Wu Yi. Xue. Gong. Cheng Xue. Za Zhi., vol. 17, no. 4, pp. 451453,
Dec. 2000.
C. D. Swerdlow, M. L. Brown, K. Lurie, J. Zhang, N. M. Wood, W. H.
Olson, and J. M. Gillberg, Discrimination of ventricular tachycardia
from supraventricular tachycardia by a downloaded wavelet-transform
morphology algorithm: A paradigm for development of implantable cardioverter defibrillator detection algorithms, J. Cardiovasc. Electrophysiol., vol. 13, no. 5, pp. 432441, May 2002.
J. P. Martinez, R. Almeida, S. Olmos, A. P. Rocha, and P. Laguna, A
wavelet-based ECG delineator: Evaluation on standard databases, IEEE
Trans. Biomed. Eng., vol. 51, no. 4, pp. 570581, Apr. 2004.
A. Van Oosterom, Solidifying the solid angle, J. Electrocardiol., vol. 35,
pp. 181192, 2002.

1398

[28] S. Mallat, A theory for multiresolution signal decomposition: The wavelet

representation, IEEE Pattern Anal. Mach. Intell., vol. 11, no. 7, pp. 674
693, Jan. 1989.
[29] I. Chouvarda, N. Maglaveras, C. Pappas, F. J. van Capelle, and
J. DeBakker, Estimation of distance between a unipolar recording electrode and a myocardial bundle based on signal characteristics, Ann.
Biomed. Eng., vol. 32, no. 10, pp. 13361347, Oct. 2004.
[30] G. W. Botteron and J. M. Smith, Quantitative assessment of the spatial
organization of atrial fibrillation in the intact human heart, Circulation,
vol. 93, no. 3, pp. 513518, Feb. 1996.
[31] F. Ng and J. Goldberger, Understanding and interpreting dominant frequency analysis of AF electrograms, J Cardiovasc. Electrophysiol.,
vol. 18, pp. 680685, 2007.
[32] A. C. Skanes, R. Mandapati, O. Berenfeld, J. M. Davidenko, and J. Jalife,
Spatiotemporal periodicity during atrial fibrillation in the isolated sheep
heart, Circulation, vol. 98, no. 12, pp. 12361248, Sep. 1998.
[33] M. Mansour, R. Mandapati, O. Berenfeld, J. Chen, F. H. Samie, and
J. Jalife, Left-to-right gradient of atrial frequencies during acute atrial
fibrillation in the isolated sheep heart, Circulation, vol. 103, no. 21,
pp. 26312636, May 2001.
[34] S. Lazar, S. Dixit, F. E. Marchlinski, D. J. Callans, and E. P. Gerstenfeld,
Presence of left-to-right atrial frequency gradient in paroxysmal but not
persistent atrial fibrillation in humans, Circulation, vol. 110, no. 20,
pp. 31813186, Nov. 2004.
[35] J. Sahadevan, K. Ryu, L. Peltz, C. M. Khrestian, R. W. Stewart, A. H.
Markowitz, and A. L. Waldo, Epicardial mapping of chronic atrial fibrillation in patients: Preliminary observations, Circulation, vol. 110, no. 21,
pp. 32933299, Nov. 2004.
[36] L. Faes and F. Ravelli, A morphology-based approach to the evaluation
of atrial fibrillation organization, IEEE Eng. Med. Biol. Mag., vol. 26,
no. 4, pp. 5967, Jul. 2007.
[37] K. Nademanee, J. McKenzie, E. Kosar, M. Schwab, B. Sunsaneewitayakul,
T. Vasavakul, C. Khunnawat, and T. Ngarmukos, A new approach for
catheter ablation of atrial fibrillation: Mapping of the electrophysiologic
substrate, J. Amer. Coll. Cardiol., vol. 43, no. 11, pp. 20442053, Jun.
2004.
[38] Y. Takahashi, M. D. ONeill, M. Hocini, R. Dubois, S. Matsuo, S. Knecht,
S. Mahapatra, K. T. Lim, P. Jais, A. Jonsson, F. Sacher, P. Sanders,
T. Rostock, P. Bordachar, J. Clementy, G. J. Klein, and M. Haissaguerre,
Characterization of electrograms associated with termination of chronic
atrial fibrillation by catheter ablation, J. Amer. Coll. Cardiol., vol. 51,
no. 10, pp. 10031010, Mar. 2008.
[39] K. Nademanee, M. C. Schwab, E. M. Kosar, M. Karwecki, M. D. Moran,
N. Visessook, A. D. Michael, and T. Ngarmukos, Clinical outcomes of
catheter substrate ablation for high-risk patients with atrial fibrillation,
J. Amer. Coll. Cardiol., vol. 51, no. 8, pp. 843849, Feb. 2008.
[40] D. Scherr, D. Dalal, A. Cheema, S. Nazarian, I. Almasry, K. Bilchick,
A. Cheng, C. A. Henrikson, D. Spragg, J. E. Marine, R. D. Berger,
H. Calkins, and J. Dong, Long- and short-term temporal stability of complex fractionated atrial electrograms in human left atrium during atrial
fibrillation, J. Cardiovasc. Electrophysiol., vol. 20, no. 1, pp. 1321,
Jan. 2009.
[41] M. Porter, W. Spear, J. G. Akar, R. Helms, N. Brysiewicz, P. Santucci, and
D. J. Wilber, Prospective study of atrial fibrillation termination during
ablation guided by automated detection of fractionated electrograms,
J. Cardiovasc. Electrophysiol., vol. 19, no. 6, pp. 613620, Jun. 2008.

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 6, JUNE 2010

Richard P. M. Houben (M05SM06) received the

B.S. degree (with highest honors) in electrical engineering from the Technische Hogeschool, Heerlen,
The Netherlands, in 1984.
From 1984 to 1989, he was engaged in industrial research and development of ultrasound imaging systems and the analysis of coronary angiograms.
Since 1989, he has been a Scientist at the Medtronic
Bakken Research Center, Research and Technology,
Maastricht, The Netherlands. His current research interests include processing of biomedical signals, especially focusing on the analysis of ECG and hemodynamic parameters in heart
failure, and spatiotemporal analysis of atrial fibrillation.

Natasja M. S. de Groot received the Medical degree

from Leiden University Medical School, Leiden,
The Netherlands, in 1998, and the Ph.D. degree in
2006 from Maastricht University, Maastricht, The
Netherlands.
She finished her training in cardiology in 2008.
She is currently a Cardiologist/Electrophysiologist
at Erasmus Medical Center, Rotterdam, The
Netherlands. Her research interests include mapping
studies of atrial fibrillation and atrial tachyarrhythmias in patients with congenital heart disease.

Maurits A. Allessie received the Ph.D. degree

from the University of Amsterdam, Amsterdam, The
Netherlands, in 1977.
In 1984, he was appointed a Professor of physiology at Maastricht University, Maastricht, The
Netherlands, where he is currently in the Department
of Physiology, Cardiovascular Research Institute
Maastricht.
Prof. Allessie is a member of the Editorial Board
of the Journal of Cardiovascular Electrophysiology and Cardiac Electrophysiology Review, a Guest
Editor of the Cardiovascular Research, and a member of the Royal Netherlands
Academy of Arts and Sciences, Koninklijke Nederlandse Akademie van Wetenschappen (KNAW). He coordinates experimental studies on electrophysiologic
mechanisms of cardiac arrhythmias, especially atrial fibrillation. In 2003, he
was appointed Academy Professor by KNAW.