-
Analysis of RS-Segment to Evaluate the Eect of
VentricularDepolarization during Ischemia
Akash Kumar Bhoi1,, Karma Sonam Sherpa2 and Sushant Konar1
1Department AE&I Engg., 2Department E&E Engg.,Sikkim
Manipal Institute of Technology (SMIT), Majhitar.
e-mail: [email protected]
Abstract. The QRS changes during ischemia have historically been
more dicult to parameterizeand have not come into clinical
practice. This paper presented a new approach to analyze ischemiaby
time parameter extraction of RS-Segment of the QRS complex. The
proposed methodologymainly focused on two prominent areas; rst:
detection of R and S points via Fast Fourier Transform(FFT) based
windowing & thresholding techniques with a sliding edge method.
Second: calculatingthe RS-Duration. The performances of the
detection methods are validated and RS-Duration isevaluated with
the Fantasia database (Fantasia) for 20 healthy subjects &
Long-Term ST Database(LTSTDB) for 80 ischemic patients. The
RS-Segment detection sensitivity (Se) and specicity (Sp)are
calculated 100% for Fantasia Database, whereas sensitivity (Se) is
91.6% and specicity (Sp) is974% for LTSTDB.
Keywords: Ischemia, RS-Segment, Fast fourier transform,
Windowing, thresholding, Sliding edgemethod, FANTASIA, LTSTDB.
1. Introduction
Myocardial ischemia causes changes in the STT wave, but unlike a
full thickness myocardial infarctionhas no direct eects on the QRS
complex [4]. Computer-generated X-Y plots were used to examine
thecorrelations between the magnitude of S-T depression and the R
wave and total RS amplitudes [14].In [6], ischemic changes in the
electrocardiogram (ECG) may precede angina pain. In [8], wideningof
the QRS complex or amplitude change and QRS slope information help
in myocardial ischemiadetection.
Recent studies have suggested that a decrease in high-frequency
content (150250Hz) of the QRScomplex is a better marker of ischemia
than the traditional ST index [1012]. Early animal
studiesdemonstrated changes in QRS morphology due to slowing of
intra-myocardial conduction during ischemia[911]. In [13], RMS
voltage reduction of the high-frequency QRS components (HFQRS)
presents largeinter individual variation, making this index
incompetent for separation of subjects with and withoutcoronary
artery disease (CAD) & MI [13,1]. The R-waves of ECG are
detected using slope detectiontechnique and proper thresholding
[2]. Calculation of Instantaneous Heart Rate (IHR) described in
[3].
Corresponding author.
ICC-2014 Editors: K. R. Venugopal and A. C. Ramachandra pp.
105111. 105
-
Akash Kumar Bhoi et al.
Figure 1. ECG inection points.
In a number of epidemiological studies involving ventricular
repolarization abnormalities in the electro-cardiogram (T wave
morphology and QT interval prolongation) related with Sudden
Cardiac Death [15]and of cardiovascular death, [16] probably
because they could be markers of ventricular hypertrophy,
leftventricular dysfunction or myocardial ischemia [17]. In [7],
authors proposed HamiltonTompkins andHilbert transform-based
methods for QRS detection and modied threshold method. Evaluation
of QTintervals for acute myocardial ischemia analysis proposed in
[18]. QRS morphology is used for the pur-pose of cardiac
arrhythmias diagnosis, conduction abnormalities, ventricular
hypertrophy, myocardialinfarction, electrolyte derangements, etc.
[19]. Changes during depolarization phase (the QRS complex)of the
ECG also add information regarding ischemia [20].
The Proposed methodology is applied to analyze these changes and
establish a relationship betweenthe eects of RS-Segment (Figure 1)
during myocardial ischemia. The detection performance is
validatedwith Fantasia & LTSTDB databases.
1.1 The QRS Complex Features
The QRS complex is an important part of the ECG signal carrying
a number of clinically signicantparameters of cardiac arrhythmia.
The duration of QRS is one of the main characteristics of this
complexand can be used in analysis and classication of the ECG
signal. This parameter is dened as the timeit takes for
depolarization of the ventricles [5].
Based on the medical denition [21], QRS interval is the duration
between the onset and the oset ofthe QRS complex. Its normal
duration is 0.040.11s (i.e., 1540 sampling points at a sampling
frequencyof 360Hz). Tape #103 in the MITBIH arrhythmia database is
an example [22], taken for this study andthe duration time is 20
sampling points, where the distance from point Q to point R is 11
(255266)sampling points, and point R to point S is 9 (266275)
sampling points, respectively. Previously discussedmethods
suggests, the major ischemia detection being carried out by
analyzing ST-segment and T-wavewhich is the continuation of
RS-Segment. The proposed methodology is focused on formulating
cohesivelink between ischemia and RS-Duration. The performance
measurement and evaluation of the detectiontechniques are veried
with the LTSTDB and FANTASIA databases.
106
-
Analysis of RS-Segment to Evaluate the Eect of Ventricular
Depolarization during Ischemia
2. Methodology
2.1 ECG Database
The Long-Term ST Database contains 86 ECG recordings of 80 human
subjects (s20011m, s20021m . . .s30801m) [23], selected to present
dierent subject data for ST-Segment changes. The
proposedmethodologies have been tested over 2 rows (signals) and
2500 columns (samples/signal) with theduration of 10 sec having
sampling frequency: 250Hz & sampling interval: 0.004 sec [24].
Fantasiarecords f1y01, f1y02 . . . f1y10 and f2y01, f2y02 . . .
f2y10) were obtained from the young cohort, andrecords f1o01, f1o02
. . . f1o10 and f2o01, f2o02 . . . f2o10) were obtained from the
elderly cohort. Eachgroup of subjects includes equal numbers of men
and women [25]. These two dierent sets of databases(i.e. ischemic
& healthy patients data) are collected for the evaluation of
the proposed methodology.
2.2 Fast Fourier Transform (FFT)
Fourier transform is an integral of the form [26]:
F (u) =
f(x)ei2uxdx (1)
ei = cos() + i sin() (2)For sampled function continuous
transform (1) turns into discrete form [26]:
Fn =N1k=0
fkei 2N kn (3)
2.3 Inverse Fourier Transform
Expression for inverse Fourier transform is
f(x) =
F (u)ei2uxdu (4)
and its discrete counterpart is
fk =1N
N1n=0
Fnei 2N kn (5)
2.4 RS-Segment Detection
Following steps are being implemented on the ECG signals to
Detect R and S points.
FFT based windowing & thresholding techniques
Step-1: FFTRemoval of low-frequencies component
Step-2: IFFTRestoration of ECG signal
107
-
Akash Kumar Bhoi et al.
Figure 2. RS-Segment detection of LTSTDB patient data
s20071m.
Figure 3. RS-Segment detection of FANTASIA patient data
f2y09m.
Figure 4. Localization of RS-Segment for single ECG waveform (of
FANTASIA data f2y09m) for bettervisualization.
Step-3: Windowed lter (default size)Localization of maxima (only
maximum in his window and ignores all other values)
Step-4: Threshold FilterRemove small values and preserve
signicant ones
Step-5: Repeat Step-3 with adjusting size of the windowed lter
to improve ltering performance.
Step-6: R-peaks detected
Sliding edge method
Step-7: Find the sample value of R-peaks(Let x be the sample
value of detected R-peaks)
Step-8: Sliding edge method to detect S pointsCompare x & x
+ 1 and detect the value of S if, x < x + 1
Step-9: Finally, RS-Segment detected (shown in Figure 2, 3 and
4).
108
-
Analysis of RS-Segment to Evaluate the Eect of Ventricular
Depolarization during Ischemia
Table 1. RS-Duration of all subjects from LTSTDB and FANTASIA
databases.
2.5 RS-Duration Calculation
The LTSTDB signals are having length of 10 sec (2500 samples).
RS-Duration is calculated for theselected initial waveform (i.e.
500 samples or 2 sec data) of the full length signal. The length of
thedetected RS-Segment is multiplied with sampling interval (i.e.
0.004 sec) to obtain the RS duration.Table 1 shows the calculated
RS duration of each ECG signals of Fantasia and LTSTDB.
3. Results Analysis
The performance of the methodologies is evaluated by the
sensitivity (Se) and the specicity (Sp). The Seand Sp are normally
computed by:
Se = 1 FNTP + FN
=TP
TP + FN(6)
109
-
Akash Kumar Bhoi et al.
Sp = 1 FPTP + FP
=TP
TP + FP(7)
False positive (FP) indicates that the algorithm detects a beat
when no beat is present; whereas, a Falsenegative (FN) indicates
that the algorithm failed to detect a real beat. TP (true positive)
stands for thebeat, properly detected [19]. The sensitivity (Se) =
100% and specicity (Sp) = 100% for the FantasiaDatabase and the
LTSTDB results are Se = 91.6% and Sp = 97.4%. The FP and FN are
highlighted onthe Table 1. The idea of calculating RS-Duration for
both the Fantasia and LTSTDB is to project theeect on healthy and
ischemic ECG signal during ventricular depolarization. The mean
RS-Duration ofFantasia and LTSTDB are 0.0323 sec & 0.0430 sec,
respectively. The dierence in the RS-Durations ofnormal and
ischemic signals shows the abnormal changes in RS-Duration during
ischemia. The variationin RS-Duration can be observed in LTSTDB
(Table 1) where, minimum RS-Duration is found to be0.0200 sec
(s20131m, ML2 lead) and maximum is 0.1120 sec (s30671m, V6
lead).
4. Conclusion
The previously proposed scheme uses various stages, including
pre-processing, conditioning and postprocessing. The implemented
methodologies have overcome such pre-processing stages. New
methodo-logies have been discussed for parameterization of
RS-Segment with optimum results. Se and Sp are foundto be 100% for
healthy signals (i.e. FANTASIA) and 91.6% & 97.4% for patients
with ischemic conditions(i.e. LTSTDB) respectively. The mean
RS-Duration dierence between FANTASIA and LTSTDB is0.0107 sec,
which shows widening of RS-Segment during ischemia. This analysis
will help the academiccommunity and researchers to explore the
further ndings in the ischemic heart diseases.
References
[1] M. Ringborn et al., The absence of high-frequency QRS
changes in the presence of standard electrocardio-graphic QRS
changes of old myocardial infarction, Amer. Heart J., vol. 141, no.
4, pp. 573579, (2001).
[2] M. A. Khayer et al., ECG peak detection using wavelet
transform, 3rd International Conference onElectrical & Computer
Engineering (ICECE) 2004, 2830, Dhaka, Bangladesh, December
(2004).
[3] I. Daubechies, Ten lectures on wavelets, Society for Ind.
and App. Math. Philadelphia, Pennsylvania.ISBN 0 89871-274-2,
(1992).
[4] Kevin Channer et al., ABC of clinical electrocardiography
Myocardial ischemia, BMJ, vol. 324, April(2002).
[5] H. Gholam Hosseini et al., Ecient features for Ann-based ECG
classiers, IEEE Bio., (1999).[6] E. Pueyo et al., Time course of
ECG Depolarization and Repolarization Changes during Ischemia in
PTCA
Recordings, Methods Inf Med., vol. 43, pp. 436, (2004).[7]
Natalia M. Arzeno et al., Analysis of First-Derivative Based QRS
Detection Algorithms, IEEE Trans-
actions on Biomedical Engineering, vol. 55, no. 2, February
(2008).[8] Esther Pueyo et al., QRS Slopes for Detection and
Characterization of Myocardial Ischemia, IEEE Trans-
actions on Biomedical Engineering, vol. 55, no. 2, February
(2008).[9] R. P. Holland and H. Brooks, The QRS complex during
myocardial ischemia: An experimental analysis in
the porcine heart, J. Clin. Invest., vol. 57, pp. 541550,
(1976).[10] J. Pettersson et al., Changes in high-frequency QRS
components are more sensitive than ST-segment
deviation for detecting acute coronary artery occlusion, J.
Amer. Coll. Cardiol., vol. 36, pp. 18271834,(2000).
[11] R. L. Hamlin et al., QRS alterations immediately following
production of left ventricular freewall ischemiain dogs, 4m. J.
Physiol., vol. 215, pp. 10321040, (1968).
[12] A. Beker, et al., Analysis of high frequency QRS potential
during exercise testing in patients with coronaryartery disease and
in healthy subjects, Pacing Clin. Electrophysiol., vol. 19, no. 12,
pp. 20402050, (1996).
110
-
Analysis of RS-Segment to Evaluate the Eect of Ventricular
Depolarization during Ischemia
[13] S. Abboud, et al., High frequency electrocardiography using
an advanced method of signal averaging fornon-invasive detection of
coronary artery disease in patients with normal conventional
electrocardiogram,J. Electrocardiol., vol. 19, no. 4, pp. 371380,
(1986).
[14] M. C. Gerson et al., Relation of exercise-induced
physiologic S-T segment depression to R wave amplitudein normal
subjects, Am J. Cardiol., (1980).
[15] S. M. Straus, J. A. Kors, M. L. de Bruin, C. S. van der
Hooft, A. Hofman, J. Heeringa et al., ProlongedQTc interval and
risk of sudden cardiac death in a population of older adults, J. Am
Coll. Cardiol., vol. 47,pp. 3627, (2006).
[16] D. De Bacquer, G. de Backer, M. Kornitzer and H. Blackburn,
Prognostic value of ECG ndings for total,cardiovascular disease,
and coronary heart disease death in men and women, Heart, vol. 80,
pp. 5707,(1998).
[17] M. H. Lehmann and F. Morady, QT interval: metric for
cardiac prognosis, Am. J. Med., vol. 115,pp. 7324, (2003).
[18] Javier Jimenez-Candil et al., QT Interval and Acute
Myocardial Ischemia: Past Promises, New Evidences,Rev. Esp.
Cardiol., vol. 61(6), pp. 5613, (2008).
[19] Zine-Eddine Hadj Slimane et al., QRS complex detection
using Empirical Mode Decomposition, Else-vier Inc. All rights
reserved, (2009).
[20] D. Romero et al., A Vectorial Approach for Evaluation of
Depolarization Changes during Acute MyocardialIschemia, Computing
in Cardiology, vol. 37, pp. 265268, (2010).
[21] R. M. Rangayyan, Biomedical Signal Analysis: A Case-study
Approach, WileyInterscience, New York,pp. 1828, (2001).
[22] Yun-Chi Yeh et al., QRS complexes detection for ECG signal:
The Dierence Operation Method,Computer methods and programs in
biomedicine 91 pp. 245254, Elsevier Ireland Ltd., All rights
reserved(2008).
[23] A. L. Goldberger et al., PhysioBank, PhysioToolkit, and
PhysioNet: Components of a New ResearchResource for Complex
Physiologic Signals, Circulation 101(23): e215-e220 [Circulation
Electronic
Pages;http://circ.ahajournals.org/cgi/content/full/101/23/e215],
June (2000).
[24] Franc Jager et al., Long-term ST database: a reference for
the development and evaluation of automatedischaemia detectors and
for the study of the dynamics of myocardial ischaemia, Medical
& BiologicalEngineering & Computing, vol. 41(2), pp.
172183, (2003).
[25] N. Iyengar et al., Age-related alterations in the fractal
scaling of cardiac interbeat interval dynamics,Am. J. Physiol.,
vol. 271, pp. 10781084, (1996).
[26] Mohamed Elgendi, Fast QRS Detection with an Optimized
Knowledge-Based Method: Evaluation on 11Standard ECG Databases,
PLoS One. 2013; 8(9): e73557. doi: 10.1371/journal.pone.0073557,
Sep. (2013).
111
/ColorImageDict > /JPEG2000ColorACSImageDict >
/JPEG2000ColorImageDict > /AntiAliasGrayImages false
/CropGrayImages true /GrayImageMinResolution 300
/GrayImageMinResolutionPolicy /OK /DownsampleGrayImages false
/GrayImageDownsampleType /Bicubic /GrayImageResolution 300
/GrayImageDepth 8 /GrayImageMinDownsampleDepth 2
/GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true
/GrayImageFilter /FlateEncode /AutoFilterGrayImages false
/GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict >
/GrayImageDict > /JPEG2000GrayACSImageDict >
/JPEG2000GrayImageDict > /AntiAliasMonoImages false
/CropMonoImages true /MonoImageMinResolution 1200
/MonoImageMinResolutionPolicy /OK /DownsampleMonoImages false
/MonoImageDownsampleType /Bicubic /MonoImageResolution 1200
/MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000
/EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode
/MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None
] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false
/PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000
0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true
/PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ]
/PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier ()
/PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped
/False
/Description > /Namespace [ (Adobe) (Common) (1.0) ]
/OtherNamespaces [ > /FormElements false /GenerateStructure
false /IncludeBookmarks false /IncludeHyperlinks false
/IncludeInteractive false /IncludeLayers false /IncludeProfiles
false /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe)
(CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector
/DocumentCMYK /PreserveEditing true /UntaggedCMYKHandling
/LeaveUntagged /UntaggedRGBHandling /UseDocumentProfile
/UseDocumentBleed false >> ]>> setdistillerparams>
setpagedevice