Arrhythmia analysis by successive RR plotting

Arrhythmia

Analysis by Successive RR Plotting

Tsuyoshi Anan, MD, Kenji Sunagawa, MD, Haruo Araki, MD, and Motoomi Nakamura, MD

Abstract: A successive RR interval plot was developed to analyze arrhythmia. The plot consisted of a set of points with the x-value of (N)th RR interval and the y-value of (N + I)th RR interval. This method was applied in the arrhythmia analysis of Holter electrocardiograms obtained from 35 patients. In the analysis of ventricular premature contractions (VPCs) this method was useful not only in detecting VPCs but also in demonstrating coupling interval-dependent characteristics of VPCs. In the analysis of atria1 fibrillation the successive RR plot enabled the authors to estimate the functional refractory period of the atrioventricular conduction. In conclusion, despite its simplicity, the successive RR piot was found to be powerful in analyzing arrhythmia. Specifically, the potential to analyze integrally the coupling interval-dependent properties of various types of arrhythmia makes it attractive as a clinical tool. Key words: arrhythmia, ventricular premature contractions, atria1 fibrillation, refractory period.

rithms. If we are able to make use of the coupling interval-dependent properties of various types of arrhythmia by analyzing them over many beats, we may uncover their hidden characteristics, which are not recognizable otherwise. Therefore, the purpose of this investigation is to develop a new technique to diagnose arrhy~mia by integrally analyzing the coupling interval-dependent characteristics of arrhythmia. To achieve this goal, we developed the successive RR interval plot. We have demonstrated that the plot is useful both in detecting and in highlighting specific features of various types of arrhythmia. Illustrated in Figure 1 is the basic principle of the sequential RR interval plotting.3-5 Each point has the x-value of the (N)th R-to-R interval (RR interval) and the y-value of (N + 1)th RR interval. Sequential plotting of these points yields the successive RR plot. In reference to Figure 1, the first two RR intervals are constant (ie, RR interval = a). Therefore, we

The long-term electrocardiographic recording has been known to be useful for the diagnosis of arrhythmia. Since massive data must be analyzed, major efforts have been made to develop efficient computer algorithms. Central to the accurate diagnosis of arrhythmia is the recognition of disturbed rhythm. Conventional algorit~s identify arrhythmia through scanning of the beat-to-beat periodicity. ‘J Even though the preceding coupling interval is known to play a crucial role in the occurrence of arrhythmia, this property in general has not been taken into consideration in the conventional algoFrom the Research institute of Ang~~ard~ology and ~ardiovas~lar Clitic, Kyushu ~~~versi~ Medical School, Fukuoka, Japan.

Supported in part by Research Grant for Cardiovascular Diseases from the Ministry of Health and Welfare, and by Grandin-Aid for General Scientific Research (01570493) from the Ministry of Education, Science and Culture of Japan. Reprint requests: Kenji Sunagawa, MD, Research Institute of Angiocardiology and Cardiovascular Clinic, Kyushu University Medical School, 3- l- 1 Maidashi, Higashiku, Fukuoka 8 12, Japan. (59C-5)

243

244

Journal of Electrocardiology

Vol. 23 No. 3 July 1990

Materials and Methods

L,I_,Jb-L-Lal-a

J

kec > C

_

a

I? %

b

a

b RR n

C

(SK

I

Fig. 1. Successive RR plot of VF’Cwith a full compensatory

pause. The abscissa is the (n) -th RR interval and the ordinate is the (n + 1 distribution labeled A on the diagonal line represents the basic regular rhythm. Any points below the diagonal line have the shorter coupling interval relative to the preceding RR interval.

obtain point A, which appears directly on the diagonal line. The third RR interval (ie, RR interval = b) reflects Because of this the point moves below the diagonal line to 8. The fourth RR interval has a pause (ie, RR interval = c). The of the coupling interval moves the point to 6. The fifth RR interval returns to the initial value. This moves the point to D. Since the fifth and sixth interval are identical, the last point returns to A. This simple example illustrates fundamental characteristics of the successive RR plot. When the basic rhythm is reasonably regular, points them distribute along the diagonal line. Any points below the diagonal line indicate a shorter coupling interval relative to the preceding intervals. By definition, therefore, any points below the diagonal line represent either atria1 premature or ventricular premature (VPCs) .

We analyzed the long-term electrocardiograms (ECGs) obtained from 20 patients with a variety of VPCs including parasystole, and from 15 patients with atria1 fibrillation, 5 of whom had accessory pathways. We recorded these long-term ECGs using a portable tape recorder (Avionics, Model 445) and played them back with a dedicated scanner (Avionics, Model 660A) at a rate 60 times faster than the actual rate. Only a single lead was used for the subsequent analyses. The output signal from the scanner was preprocessed with home-wired hardware, which consisted of a bandpass filter for noise reduction and an analog differentiator, to ensure the recognition of QRS complexes. The passband was set between 0.5 and 60 Hz. The time constant of the differentiator was carefully chosen so that the R wave was accurately recognized. These preprocessed signals (ie, noise-reduced original signal and its firsttime derivative) were digitized at 10 kHz, which is equivalent to 167 Hz in the original time scale, with a laboratory computer system (Mitsubishi, Melcom 70). A leading edge of the QRS complex was recognized when both ECG and its time derivative simultaneous~y satisfied a prespecified criterion. We could identify up to 16 different waveforms of QRS complex by a template-matching technique.’

Results Ventricular Premature Contraction The successive RR plots of VPCs varied in their basic rhythms and coupling intervals. Figure 2A is an example of VPCs with a fixed coupling interval. The dense distribution labeled A, in the area of the diagonal line, indicates the regular basic rhythm (ie, sinus rhythm). The group of points labeled B represents VPCs. Constancy of the coupling interval, irrespective of the basic rhythm, manifests as the narrow distribution parallel to the x-axis. The maximal variability of the coupling interval was less than 120 msec. The group of points labeled C, distributed above the diagonal line, reflects the apparent compensatory pause following the preceding VPC. The dis~butions represented by A, B, C, and D in Figure 2 indicate the characteristic pattern of the successive RR plot of the fixed coupling VPCs with a complete compensatory pause.

Successive RR Plot

(secl

l

Anan et al.

245

-

1.32.

_,

C

0.84.

0.84.

0.35,

0.36.

2

r

1

I.‘.“. 0

0.36

0.84

RRil

1.32

0

I 0.36

A

0.84

RRn

0

1.32

“’ 0.36

I3

0.84

RRn

1.32

(set 1

C

Fig. 2. (A) VPC with futed coupling interval. The abscissa is the n-th RR interval and the ordinate

is the (n + l)-th RR interval. Distribution A, in the area of the diagonal line, indicates the regular basic rhythm; distribution B, below the diagonal line, represents VPCs; and distribution C, above the diagonal line, shows postmature contractions. (B) Variable coupling VPCs. Compared to (A), each of the four distributions (A, B, C, and D) is widely scattered. (C) Ventricular parasystole. Vertical distribution B represents significant variability in the coupling interval, which varies between 0.38 and 0.80 sec.

Figure

2B shows

VPCs with a variable coupling

interval. Compared to Figure 2A, each one of the four distributions (A, B, C, and D in Fig. 2B) is considerably scattered. Implicit to this observation is the result that there are sizable variabilities in the basic rhythm, the coupling interval of VPCs, and the compensatory pause. Illustrated in Figure 2C is ventricular parasystole. The coupling interval varies as evidenced by the nearly vertical distribution labeled B. The ventricular refractoriness constitutes the shortest coupling interval, while the RR interval of the basic rhythm constitutes the longest coupling interval. Although all three VPCs were confirmed to have the full compensatory pause, the distribution of the successive RR plot is remarkably different.

Atrial Fibrillation Figure 3 shows the successive RR plot of common atria1 fibrillation obtained from a 20-minute recording. Points were scattered widely and were no longer distributed along the diagonal line. This implies that the variability of the coupling interval (ie, RR interval) was increased. This distribution was very sensitive and specific and thus diagnostic for all atria1 fibrillation subjected to the analysis in this investigation. The y-value of the envelope of the successive RR

plot along the x-axis (ie, the thick dashed curvilinear line in Fig. 3A) increased as the x-value increased. Since the x-value represents the preceding cycle length, the y-value of the envelope indicates the shortest attainable coupling interval as a function of the preceding cycle length (ie, basic cycle length). The functional refractory period of a tissue is defined’ as “the shortest attainable interval between two impulses, the basic and the premature, transversing that tissue and measured at a point distal to the tissue.” If we accept this definition, the envelope represents the functional refractory period of the AV conduction as long as the frequency of repetitive stimulation to the AV node in atria1 fibrillation is faster than its refractoriness. Depending upon the rate of ventricular response, the length of recording required to estimate the envelope was different. For the heart with a slow ventricular response, a rather longer recording time was required for the accurate estimation and vice versa. In general, a 20-minute recording will suffice. Shown in Figure 3B is Wolff-Parkinson-White (WPW) syndrome with atria1 fibrillation. Unlike common atria1 fibrillation (Fig. 3A), the y-value of the envelope (ie, the thick dashed line) remains virtually constant (230 msec), irrespective of the xvalue. This demonstrates that the functional refractory period of the AV conduction was insensitive to the preceding cycle length. Whether this is a unique characteristic of the accessory bundle remains to be seen.’

246

Journal of Electrocardiology

Vol. 23 No. 3 July 1990

(set It 1.32 i

0.36-

0

0.84

0.36

RRn Fig. 3.

RR

1.32

0

0.84

0.36

1.32

RRn

A

plot of

observation was diagnostic for all atria1 fibrillations analyzed. The thick dashed line represents the functional refractory period of the AV Note that it varies with the RR interval. (B) Successive RR plot of WPW syndrome with atria1 fibrillation. The thick dashed line indicates the functional refractory period of the accessory pathway. Note that the refractory period of the accessory pathway remains remarkably constant irrespective

Premature Contractions

arrhythmia. In the analysis of VPCs, we recognized wide QRS complexes by the waveform criteria and the disturbed rhythm by the successive RR plot. This combined analysis made it possible to differentiate a variety of VPCs, such as fixed coupling VPCs, variable coupling VPCs, and parasystole, and significantly improved the efficiency of of such cycle length-dependent

where RR is the RR interval. Therefore, once we know the QT,, changes in QT, with changes in RR are predictable. curvilinear line in Figure 4 represents QT, as a function of the RR interval. Any VPCs below this line suggest that the R on T phenomenon is taking place. This graphical analysis provides an immediate representation

Fibrillation The rhythm analysis by the successive RR plot provided the definitive diagnosis of atria1 fibrillation

threatening This phenomenon is termed “R on T.” If we accept Bazett’s empirical formula,9 which relates the QT interval (QT,) to corrected QT interval (QT,), QT, is expressed as QT, = QT, ERR

conventional extra technique to determine functional refractory period, we first pace the atrium with a basic cycle length for several beats, then impose test stimuli with an interval. In the successive RR plot, however, there are no

Successive RR Plot

l

Anan et al.

247

1.44

QTc

0.72

0.74

I

034

I

t

0.4%

,

t

0.72

i

0.96

I

I

1.20

i

*

1.44

*

1

1

168

( set ) of R on phenomenon. The abscissa is RR interval is the (n + I)-th RR interval. The solid curvilinear line represents QT, as a function of the preceding RR interval. Any VPCs below this line manifest the R on T phenomenon. RRn

Fig. 4. Successive

priming beats. every beat a difpreceding cycle Nevertheless, the velope of successive RR indicates the attainable coupling for a preceding cycle This is, definition, the refractory of the conduction. The refractory period becomes shorter the precycle length basic cycle is shortThis observation qualitatively consistent the characteristics the functionai period of AV node determined by techniques.‘“-14 Quantitative of the refractory period by the cessive RR plot and techniques remains be seen. The AV refractory period of common atria1 fibrillation varied with the preceding RR interval (Fig. 3A), whereas that of WPW syndrome remained virtually constant (Fig. 3B). Although further investigations are obviously needed to understand the pathophysiological significance of the difference of the refractoriness between them, the difference was rec-

ognizable only through the RR interval analysis. Once we know the refractory period of AV conduction, we may recognize true VPCs in the aberrant ventricular conductions in atria1 fibrillation. Any beats with wide QRS complexes distributed below the envelope (the thick dashed curvilinear line in Fig. 3B) indicate true VPCs because their coupling intervals are shorter than the functional refractory period of AV conduction. The same observations seen above the envelope represent either VPCs or aberrant ventricular conductions.

Limitations We have indicated that the successive RR plot has unusually great potential in analyzing cardiac rhythm. There are, however, various limitations. First, to take full advantage of the successive RR interval plot, morphological recognition of QRS is mandatory. Obviously, rhythm analyses alone by the

248 Journal of Electrocardiology

Vol. 23 No. 3 July 1990

successive RR plot would not allow us to differentiate between atria1 premature contractions and ventricular premature contractions. Therefore, appropriate QRS recognition is a most basic requirement for the rhythm analysis, even with the successive RR plot. Second, with the successive RR plot, both axes are in effect parameterized with time. Therefore, only rhythm irregularity recognizable by comparing adjacent RR intervals will be identified. Finally, for the practical application of the successive RR plot, one must have specially designed, dedicated hardware. The device we designed is capable of generating the successive RR plot. However, the accuracy of the QRS morphology is somewhat limited because we used a single lead ECG. A more elaborate device (which can be interfaced with commercially available sophisticated Holter scanners and make use of its ability to diagnose QRS morphology) will dramatically improve the accuracy of the successive RR interval plot.

Conclusions The framework of the successive RR plot is not new, as seen in the Lorenz plot.3 The Lorenz plot was used to analyze complex periodicity in random signals. The application of this plot has been limited, however, to geophysics and other disciplines. There has been no elaborate application of this method in arrhythmia analysis. As we indicate in this preliminary report, the successive RR plot offers new views in the analysis of VPC, R on T phenomenon, aberrant ventricular conduction, and refractoriness of AV conduction. Although further significance and potentials of the successive RR plot remain to be investigated, its ability to recognize the unique coupling interval-dependent characteristics of arrhythmia in the long-term ECG makes it attractive for clinical applications.

Acknowledgment The authors thank Dr. Akiko Suyama, Research Institute of Angiocardiology and Cardiovascular Clinic, Kyushu University Medical School, for valuable comments and discussions in preparing the

manuscript. The authors also thank Drs. Hideki Yamazaki and Kazumi Hirakawa, Department of Electrical Engineering, Kyushu University for their comments on this work.

References 1. Okajima M, Stark L, Whipple GH, Yasui S: Computer pattern recognition techniques, some results with real electrocardiographic data. IEEE Trans Biomed Eng BME-10:106, 1963 2. Gersh W, Eddy DM, Dong JE: Cardiac arrhythmia classification: a heart-beat interval-marker chain approach. Comput Biomed Res 4:385, 1970 3. Lorenz EN: Deterministic nonperiodic flow. J Atmospheric Sciences 10: 130, 1963 4. Anan T, Nakagaki 0, Nakamura M: An application of the Lorenz and multi-Lorenz plot to computerized analysis of arrhythmia. Jpn Heart J 23 (suppl) :694, 1982 5. Yamazaki H, Anan T, Nakagaki 0 et al: Analysis of the ECG rhythm in the two dimensional graphic representation. Transaction of the Institute of Electronics and Communication Engineers of Japan J65-D:40, 1982 6. Winter DA, Trenholm BG: Reliable triggering for exercise electrocardiograms. IEEE Trans Biomed Eng BME-16:75, 1969 7. Narula OS: His bundle electrocardiography and clinical electrophysiology. FA Davis, Philadelphia, 1975 8. Tonkin AM, Miller HC, Svenson RH et al: Refractory periods of the accessory pathway in Wolff-ParkinsonWhite syndrome. Circulation 52:563, 1975 9. Bazett HC: An analysis of the time relationship of electrocardiograms. Heart 7:353, 1975 10. Hoffman BF, Cranefield PF: Electrophysiology of the heart. McGraw-Hill, New York, 1960 11. Moe GK, Mendez C, Han J: Aberrant A-V impulse propagation in the dog heart: a study of functional bundle branch block. Circ Res 16:261, 1965 12. Merideth J, Mendez C, Mueller WJ, Moe GK: Electrical excitability of atrioventricular nodal cells. Circ Res 23:69, 1968 13. Denes P, Wu D, Dhingra R et al: The effects of CyCk length on cardiac refractory periods in man. Circulation 49~32, 1974 14. Wiener I, Kunkes S, Rubin D et al: Effect of sudden changes in cycle length on human atrial, atrioventricular nodal and ventricular refractory period. Circulation 64:245, 1981