Body surface distributions of repolarization potentials after acute myocardial infarction. III. Dipole ranging in normal subjects and in patients with acute myocardial infarction

J. ELECTROCARDIOLOGY 14 (4), 1981,387-398

Body Surface Distributions of Repolarization Potentials after Acute Myocardial Infarction. III. Dipole Ranging in Normal Subjects and in Patients with Acute Myocardial Infarction BY DAVlD M. MIRvrs, M.D. AND MARK A. HOLBROOK, M.S.

SUMMARY Dipole ranging is a method to determine the location, strength and orientation of the heart vector from body surface measurements. This technique was applied to electrocardiograhic data recorded from forty-five normal subjects and from twenty patients suffering an acute myocardial infarction. Moments and locations during the QRST interval were determined by the potential integration formulae of Gabor and Nelson. Results in the normal population showed stability of both variables during cardiac recovery. In subjects with acute infarction, dipole loci were stable during much of the ST-segments of loci compatible with the anatomic positions of the lesions. Thus, dipole ranging methods can provide physiologically relevant information when applied to human electrocardiographic problems. The ECG has provided critically valuable information as to the diagnosis, location and extent of acute myocardial infarction>", Most studies have relied upon qualitative study of standard electrocardiographic, vectorcardiographic or body surface isopotential distributions.tw' Others have been based upon simple arithmetic schemata, providing estimates of pathophysiologic parameters. 3 ,5 We report here our attempts to quantitate cardiac source characteristics from body surface recordings. The method, based upon the potential integration formulae of Gabor and Nelson," served to determine the location and moment of the putative cardiac dipole in normal subjects as well as in patients suffering an acute myocardial infarction. Results demonstrated the ability of such techniques to provide physiologically meaningful data.

MATERIALS AND METHODS Study Populations. Two groups of subjects were studied. First, forty-five persons free of cardiac disease as determined by history, physical examination and routine electrocardiography served as a normal control population. Second, twenty men suffering an acute myocardial infarction were studied. All were evaluated within forty-eight hours of the onset of symptoms while hospitalized in a coronary care unit. The diagnosis of infarction was based upon standard historical, electrocardiographic and enzymatic criteria. None were in more than mild cardiac failure, had clinical evidence of pericarditis, nor received cardioactive agents other than lidocaine within two days of study. All were in sinus rhythm without bundle branch block. Voluntary, informed consent was granted by members of both groups prior to study. Data Acquisition. ECG signals were registered from 150 silver-silver chloride electrodes fixed to the anterior (100 electrodes) and posterior (50 electrodes) thorax in ten rows extending from the level of the clavicles to inferior rib margins. Additional electrodes were placed on the extremities to derive the Wilson central terminal potential. Interelectrode distances and torso dimensions were recorded. Voltages from each electrode were amplified by low noise, differential (torso electrode voltage vs Wilson central terminal potential) amplifiers with a high frequency cutoff of 1000 hertz and a time constant of two seconds. Gains and offsets were individually set at 1000 to 16,000, and from -4.5 to 4.5 volts, respectively, under computer control so that amplifier output filled the input range of the analog-to- digital converter. Data were acquired in five sets," each consisting of

From the Section of Medical Physics, Division of Circulatory Diseases, Department of Medicine, University of Tennessee Center for the Health Sciences, Memphis, Tennessee. Supported by USPHS grant HL·20597 and Research Career Development Award HL-000560 from the National Heart, Lung and Blood Institute, National Institutes of Health, Bethesda, MD. The costsofpublication ofthis article weredefrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. § 1734 solely to indicate this fact. Reprint requests to: DavidM. Mirvis, M.D.,SectionofMedical Physics, University of Tennessee Center for the Health Sciences, 956 Court Ave., Room 2F18, Memphis, TN 38163.

387

MIRVIS AND HOLBROOK

388

Anterior

1 . . - - -.....- . .

+X

Left

Cranial

+X

Left

-Y' c

Fig. 1. Cross-sectional (top) and frontal (bottom) views of the right elliptical cylinder used to represent the torso for dipole ranging calculations. The origin of the Cartesian reference system, the directions of X, Y and Z vectors, and the definition of the angle (J are identified. the output of thirty torso electrodes. Twenty seconds of data from each set were converted to digital form at a sampling rate of 500 conversions per channel per second. Waveforms were visually examined during data acquisition to assure baseline stability and artefactfree recordings. Data Analysis. The data subsets were merged and representative cycles were averaged, as previously detailed," to yield one set of 150 unipolar thoracic waveforms . Onsets and offsets of the T-P, QRS, and ST-T intervals were manually determined from root-me ansquare (RMS) potential plots, and the average potential during the terminal fifty msec segment of the T-P interval was selected as a baseline reference level. Subsequent data processing followed two directions . First, isopotential maps were constructed at two msec intervals throughout the QRST interval. Contour lines were drawn at zero and at plus/minus 10, 20, 40, 60, 100, 200, 400, 600, and 1000 p:V levels, using a linearbilinear interpolation routine. Second, the location and moment of the single dipole best replicating the recorded sur face potentials were computed. These calculations were based upon the nine

equations derived by Gabor and Nelson6 •8 •9 which permitted inverse determination of both location and moment of a dipole by integration of potentials recorded on the surface of an irregularly shaped boundary. To facilitate computations, the torso was represented by a right elliptical cylinder (Fig. 1) whose major and minor hemiaxes (labelled "a" and "b" in Fig. 1) were determined from torso measurements and whose height C'c" in Fig. 1) equalled the length of the electrode strips (10.8 inches). The origin of the three dimensional reference system and the orientation of its components are likewise indicated in Fig. 1. The positive halves of the X Y, and Z vectors were directed to the subject's left, superiorly, and anteriorly, respectively, from an origin lying at the centroid of the ellipse at the level of the sternoclavicular joint. The three moment terms were calculated by the equations (Eq. 1) Mx=kfIv cosfsdl.dy, (Eq. 2) My= kA(Vt-Vb), and Mz=kfIVsinedLdy, (Eq. 3) where Mx, My, Mx=the X, Y, and Z components of dipole moment, in rnA-em; V=potential at any point on the thorax, in mY; e=angle between the normal to the body surface at the torso point yielding V and the + X axis (Fig. 1); k=average thoracic conductivity, based upon a resistivity of 476 ohm-scmw; dL=element of vertical length; Vt =average potential of points along the upper rim of the cylinder; Vb=average potential at points along the bottom of the cylinder; and A=cross sectional area of the elliptical cylinder. The X, Y and Z coordinates of dipole location were calculated by a least-squares solution of five equations: MxX-MyY=k{JIxVdzdY-JJyYdzdx} (Eq.4) MyY-MzZ=k{JIyVdzdx-JIzVdxdy} (Eq.5) MyX+MxY=k{JJyVdzdY +JIxVdzdx} (Eq.6) MzY+MyZ=k{JJzVdzdx+JJyVdxdy} (Eq. 7) MxZ+MzX=k{JJxVdxdY+JIzVdzdy} (Eq. 8) where X, Y, Z=X, Y and Z dipole locations; and x, y, z=X, Y and Z coordinates of electrode yielding potential V. All integrations were performed using Simpson's numerical method at four msec intervals during the QRST interval. Accuracy of calculation was confirmed using the forward method of Okadall to compute potentials on the surface of a cylinder generated by a conta ined dipole . Statistical Methods. All results were expressed as mean ± 1 standard deviation. Significance of results was tested at a 1% level. Grouping of dipole locations was accomplished using the iterative ISODATA (Iter aJ . ELECTROCARDIOLOGY, VOL. 14, NO.4, 1981

DIPOLE RANGING IN MAN

389

LOCATION

MOMENT

!

0.33 (f)

(J)

X

<{

>Fig. 2. Location and moment of the optimally determined cardiac dipole from one normal subject. Positions of the dipole, in inches from the origin, along the X, Y and Z axes are in the left panel and X, Y and Z component moments , in m.A-cm, are on the right. The horizontal axis is time during the cardiac cycle, with tho dashed vertical line separating QRS and ST-T intervals. Note 1) the wide swings in location and moment during the QRS complex and 2) the stability of these patterns during the ST-T interval.

'5~

';--

<:{

>-

-10

en x

(J)

x

«

<{

><

x -2

a -0.33 0.33 0 -0.33 0.33

(f) I

soo ms

tive Self-Organizing Data Analysis Technique) described by Ball and Hall.P

X

-c N

Normal subjects. An example of dipole location and moment tracking is presented in Fig. 2. During the first half of the QRS complex, the computed dipole moved progressively leftward. Subsequently, a sharp rightward, superior and

A.ST+40

v

8.ST+80

v

=J= = .; -

-

J. ELECTROCARDIOLOGY, VOL. 14, NO.4, 1981

0

500 ms

-0.33

30015/1.'

anterior movement occurred. Of note is the extrathoracic location of the dipole during terminal portions of cardiac excitation; positive Y-axis locations indicate a locus in the neck or head. Isopotential distributions during that time interval were nondipolar, characterized by dual maxima, as has been previously described.P Such nondipolarity is known to cause unphysiologic dipole ranging results.U Because of the common occur-

RESULTS

Fig. 3. Body surface isopotential distributions at points 40 (Panel A) and 80 (P anel B) msec into the ST-segment. Data were sensed from the same normal subject as in Fig. 2. Plus and minus signs correspond to the positions of the 150 torso electrodes, with the sign refle cting the polarity of sensed voltage. The sternal notch is indicated by the "V" and the location of the six standard precordial sites by solid squares; the left border is on the spine and the right edge is along the left parasternal region . Contour lines are at zero (overdr awn) plus and minus 10, 20, 40, 60, 100, and 200 J.tV levels. A single anterior maximum is observed with little difference in the patterns at the two instants.

X

_

-

-

390

MIRVIS AND HOLBROOK

0.5

Z

0.0

'2 .0

--=-::.. 20

40

80

% IN ST-T

80

100

20 -05

40

60

% IN STT

80

10e

Z OI~~~ 2C 4 0 ec 80 1 0 -so

renee of such extradipolar forces during cardiac activation,14-16 further consideration will be given only to the repolarization phase ofthe cardiac cycle which is more adequately modeled as a dipole .17,18 During the ST-T interval, the cardiac dipole remained relatively stationary at a position five inches below the sternoclavicular joint, and two inches to the left and one-half inch anterior to the middle of the chest. This would be expected to correspond to the anatomic position of the left ventricular free wall. A slight leftward motion occurred as repolarization progressed; horizontal and vertical positions remained virtually unchanged. Dipole moment during repolarization rose continually until the peak ofthe T-wave peak; values for the X, Y and Z components were 0.11, -0.07 and 0.36 rnA-em, respectively at that instant. These correspond to an overall peak dipole moment of 0.38 rnA-em. Orientation remained inferior (-Y), leftward (+X) and anterior (+Z) throughout. Isopotential distributions from this subject are presented in Fig. 3. Markings are described in the legend. In panel A, the pattern 40 msec into the ST-segment was characterized by left anterior maximum (218 ,."V) and posterior negativity. At 80 msec into ST-T, the pattern was nearly identical, with peak positive potential measuring 302 ,."V. Thus, both the location of extrema and the stability of patterns with time correspond to the above described dipole ranging results. Group data are presented in Figs. 4 and 5. To facilitate comparisons between subjects, the ST-T intervals were normalized to a common time-base prior to plotting. Mean and mean ±1 standard deviation values for location, moment and orientation are depicted in Fig. 4. The stability oflocations during ST-T described for a single case is

% IN ST·T

Fig. 4. Mean and mean ± 1 standard deviation confidence limits of dipole location (left panel), dipole moment (center panel) and dipole orientation (r i gh t panel) computed from 45 normal subjects. X, Y and Z components of each variable are plotted in inches (left), rnA-em (cen ter) or degrees from the axi s (left) as a function of time, expressed as a percent of the time-normalized ST-T interval. Mean and mean ± 1 standard deviation values are plotted.

likewise apparent for the entire group (left panel). After an initial movement near the J-point, mean locations moved by less than one-half inch. Standard deviations were small, measuring less than one-half inch in all projections at all instants beyond the initial portion of the ST-T interval. As described above, positions of the dipole were consistent with a left ventricular free wall locus. Group mean dipole moments (center panel) rose progressively to a peak during the T-wave. At that point, X, Y, and Z components measured 0.23±0.10, 0.17±0.12 and 0.42±0.21 rnA-cm, respectively. Overall moment was 0.53±0.20 rnA-em. Dipole orientations (right panel) showed a small (less than 10 degrees) but progressive shift to the left during the ST-T interval. Thus, the dipole during most of repolarization may be approximated as a fixed location, fixed orientation, variable moment source. Results of a cluster analysis procedure-s are presented in Fig. 5. In the left panel, the spatial distributions of dipole loci 40 msec into the STsegment of all 45 normals are depicted in three projections. The center of the cluster of all loci was located at a point 1.29 inches to the left, 5.54 inches below and 1.02 inches anterior to the origin . The summed-square error, quantitating the sum of the squares of the three dimensional distances from each point to the cluster center, was 22.02 inches.f This corresponds to a variance of 0.49 inches" and a standard deviation of only 0.70 inches. Thus, dipole loci were closely grouped near the cluster center, which approximates a point near the left ventricle. This was likewise suggested by the narrow confidence bands shown in Fig. 4 (left panel). Results computed 80 msec into ST-T were similar (F ig. 5B), supporting the previous suggestion that the repolarization dipole remained stationary; the cluster center moved by only 0.17 inches in three dimensional space. J . ELECTROCARDIOLOGY, VOL. 14, NO.4, 1981

DIPOLE RANGING IN MAN

8.80 MSEC

A.40MSEC

CLUSTER X= 1.44 in CENTER Y= -5.61 in Z= 1.04 in

CLUSTER X= 1.29 in CENTER Y= -5.54 in Z= 1.02 in

Z

2

SSO ERROR = 22.02 in2

D D

I:fl

2

D D

~

x 2

X D

IIIi

DO!

BB ~~ D

D

D

D

D

D

X

D

-6

D

2

X

Z

~D

D Dl:b

-5

t~D D

2

Z

D

·5 Dqr

DDD~!Co

-6

D D D

D

Y

~ 2

-5

D

D

d!:lm D

D

Ib

-5

0

~g

2

SSO ERROR =20.21in2

D

2

'C ~

DrfF

·6

391

DOD D

0 D

Y

Y

Df;D

-6

a

D

D D

D

Y

Fig. 5. Scattergrams depicting locations of cardiac dipoles at two time points during the ST-segment. Dipole loci (open boxes) are shown in three views of the reference system, whose axes are labeled. Distances from the origin are in inches. Centers of the cluster of positions and summed-square error are given for each instant. Note (1) the tightness of the cluster at the two instants and (2) the stability of the location of the grouping with time.

Acute myocardial infarction. Ten subjects suffered an acute anterior infarction (ST-segment elevation in leads V 1.4 with or without reciprocal changes in leads I and aVL) and a like number experienced acute inferior injury (ST-segm ent elevation in leads II, III and aVF with or without right precordial ST-segment depression). Dipole ranging results from one patient with

each type of lesion are superimposed in Fig. 6. As in normals (Fig. 2), wide swings in dipole location were observed during the QRS complex. Similarly, dipole loci were stationary during the first two-thirds of repolarization . Moment at the J-point deviated significantly from zero, in contrast to the results in normals. This injury current moment increased during the ST-segment as

LOCATION

0,1---+-------

Fig . 6. Location and moment plots , presented as in Fig. 2, for two subjects. One subject, identified by "A", suffered an acute anterior myocardial infarction. The second, labelled "I", experienced an acute inferior infarction. During the ST-segment, dipole locations and orientations were divergent in these two subjects.

MOMENT

0.3

x -0.33 0.33

Z

Z -2

J. ELECTROCARDIOLOGY, VOL. 14, NO.4, 1981

Of-+t+---,~~--;::-:t,

-0.33

392

A.ANTERIOR

MIRVIS AND HOLBROOK

v

Fig. 7. Isopotential distributions, constructed from potential sensed 40 msec into the ST-segment from the two subjects with acute myocardial infarctions whose data were also presented in Fig. 6. Markings and conventions are as in Fig. 3. Panel A: The pattern with an acute anterior lesion was characterized by an anterior maximum, corresponding with the anterior orientation of the cardiac dipole . Panel B : In contrast, the distribution with an inferior lesion is dominated by an anterior maximum; the operative dipole was posteriorly directed.

orientation of the vector remained fixed. At the end of ST-T, locations underwent major shifts, as did dipole orientation. Dipole locations and orientations differed in the two examples presented in Fig. 6. In the subject with an anterior lesion, labelled "A," ST-segment loci were similar to those described for normals. At a point 40 msec into ST-T, the dipole was located 1.7 inches to the left, 6.2 inches below and 0.4 inches anterior to the origin. The locations in the patient with an inferior lesion (labelled "I") were located to the right of and inferior to those of the subject with an anterior lesion. Locus 40 msec into the ST-segment was 0.3 inches to the left, 7.3 inches below, and 0.2 inches anterior to the origin. Whereas the dipole in example "A" was oriented anteriorly, inferiorly and leftward, that in example "I" was directed posteriorly, inferiorly and rightward during the ST-segment. Dipole moments measured 0.38 rnA-em and 0.13 rnA-em in examples "A" and "I", respectively, 40 msec into the ST-segment. These differences correlated with the isopotential body surface maps constructed for these subjects (Fig. 7). In panel A, the distribution 40 msec into the ST-segment for the subject with an anterior lesion can be seen to be characterized by left anterior maximum (peak potential = 694 J.L V) and posterior negativity. The distribution from the subject with an inferior lesion (Panel B) was

characterized by an anterior minimum (peak potential = 410 J.L V). The former pattern is consistent with a leftward, anteriorly oriented dipole, as computed above, projecting positive forces to the precordium, and the latter is consistent with posterior, rightward vector, as computed above, projecting negative forces to the upper anterior torso . Group data are presented in Figs. Band 9. Dipole moments in subjects with anterior lesions measure O.22±0.07 and O.22±O.08 rnA-em at instants 40 and 80 msec into the ST-segment, respectively. Corresponding values for patients with inferior lesions were 0 .14±0.OB and 0.16±0.09 rnA-em. At both instants, moments from those with anterior lesions significantly exceeded those of subjects with inferior lesions (p < 0.01, unpaired t-test). Dipole locations during the ST-T interval varied more than in normal subjects. In Fig. 8, the cumulative distance moved by the dipole from the J-point to each instant during the time normalized ST-T interval is shown. Mean and ±1 standard deviation confidence limits are plotted for normal and infarct subject groups. In normal persons, over half of the mean motion during the entire ST-T interval (6.8± 4.2 inches) occurred within the initial 5% of that period; subsequently motion was slow and even, with a mean change of 0.032 inches per percent of ST-T interval. In subJ . ELECTROCARDIOLOGY, VOL. 14, NO.4, 1981

393

DIPOLE RANGING IN MAN

using the Hotelling t-test. Thus, dipoles representing injury currents generated by anterior lesions were located to the left of and superior and anterior to those generated by inferior myocardial infarctions.

A. NORMAL ....---;"'

c

w20

oZ

DISCUSSION

~ 10

Einthoven, Fahr and deWaart-" first idealized the cardiac source as an electrical doubet or dipole which was fixed in location at the center of an equilateral triangle. Later, the dipole was permitted to be both eccentric and mobile, existing at various sites during different segments of the cardiac cycle.6 ,14 This single moving dipole cardiac equivalent generator model forms the basis of the "dipole ranging" method considered in this report. This technique is, in essence, a determination of the location, orientation and strength of the optimal cardiac dipole computed from volume-conductor surface potential measurements. 2 (}- 2 2 Two major advantages of this method may be identified. First, it permits quantitation of source strength in biophysical terms which are independent of volume conductor size. This is in contrast to the classical electrocardiographic registration

(j)

o

B. INFARCTION c W20

0

Z

~ 10

(j)

0 40 60 80 Ofo IN ST-T

100

Fig. 8. Plots of the mean and mean ±1 standard deviation values of the cumulative movement of the cardiac dipole during the time normalized ST-T intervals. Values derived from normal subjects are plotted in Panel A and those from patients with myocardialinfarction are in Panel B. jects with either type of infarction, the total motion of 12.6±6.4 inches significantly (p < 0.01) exceeded that of normals. Additionally, the rate of movement increased from 0.068 inches per percent in the early ST-T interval to 0.110 inches per percent during the last one-third of that interval. In Fig. 9, dipole locations 40 msec into the STsegment are projected into the reference system, in a manner similar to that of Fig. 5. The centroid of the cluster of loci derived only from subjects with anterior lesions was located 1.19±0.57 inches to left, 6.25±0.38 inches below, and 1.06±0.31 inches anterior to the origin. That of the cluster derived from patients with inferior lesions was computed to be 0.47±0.82 inches to the left, 6.69±0.49 inches below and 1.46±0.77 inches anterior to the origin. The differences between group centers were statistically significant J. ELECTROCARDIOLOGY, VOL. 14, NO.4, 1981

.......

c

o

W2

oZ ~1 o

o ANTERIOR o INFERIOR Ol-..----..---..---X 1

2

DISTANCE(inJ 2I o:<: 1 Xq ....r----'---.L..---Z -5

-5

o ~O -6

o

o

~ ~O

_Y DO ~

o

o o

Fig. 9. Scattergrams ofdipole lociderived from subjects with anterior (circles) and inferior (squares) infarctions, 40msecinto ST-T. Distancefrom the origin along each labelled axis is in inches.

394

MIRVIS AND HOLBROOK

of voltage, which represents the effects of the source acting through a variably dimensioned volume filled with resistive media. Second, the operative source can be localized. This cannot be accomplished by standard ECG or vectorcardiographic methods; indeed, the latter ideally requires a fixed location cardiac source. The computational routines to accomplish these goals were derived from the equations published by Gabor and Nelson in 1954.6 They demonstrated that strength, location and orientation of the resultant dipole of a system of sources and sinks immersed in a volume conductor could be determined from potentials recorded over the boundary. The first moment of the potential distribution gives the magnitude and orientation, while the second moment quantitates the location of the dipole. Forms of the equations applicable to animal torsos were then formulated.s" Several types of studies have documented the accuracy of the Gabor-Nelson approach. Gastonguay and Nelson-'' computed the parameters of artificial dipoles in a fluid filled dish; orientation was correct to within one degree and moment was accurate with a mean error of 4.8%. When frog gastrocnemii were placed in the bath and stimulated, the path of the computed dipole closely correlated with the expected loci of depolarization and repolarization boundaries. Results using a two-dimensional torso model were reported by Gabor and Nelson;" calculated locations were accurate to within one degree if a homogeneous medium was assumed. Errors in location and orientation increased to only 0.6 em and four degrees, respectively, if inhomogeneities representing bone and lung were added. In a three dimensional elliptical cylinder, errors of two percent between measured and calculated values were achieved. Nelson, Hodgkin, and Voukydis? subsequently applied the method to potentials sensed on the surface of a male and a female human torso model. When integrations were performed at but three horizontal levels, the correlation between actual and measured dipole location changes was very high (r=0.99). Applications of Gabor-Nelson theory to animal studies followed. Dipole moments during cardiac excitation were reported from a wide variety of species, ranging in size from frogs to horses. 2 4-2 8 A close expotential relationship (r=O.99) between peak QRS dipole moment and body or heart weight was suggested.s? ST-T dipole moments have also been computed for the rabbit'" and the horse. 26 Dipole locations during the QRS of the

pig were found to be in general agreement with cardiac excitation data by Hodgkin and associates.s" Finally, Nelson et a1.9 reported dipole moment and orientation variables during human cardiac activation. Other methods have likewise been devised to determine dipole parameters in animals and man. Geselowitzw utilized the so-called "shiftequations," defining the effects of a translocation of a dipole from the origin, to compute dipole locations. This formed the basis for localization of a dipole in man during the PQRST interval."? Lead sets sensitive to dipole and higher order multipole terms have been constructed by Schubert.F' Trost et al.,32 and Horan et al.;33 these are not, however, able to determine dipole location. Barber and Fischman-" reported a multi-electrode grid system capable of sensing total dipole moment which was used by Ellison et al. 35 in man to relate leftward oriented moment to left ventricular mass. Here too, dipole location was unavailable. Holt and associates'
DIPOLE RANGING IN MAN

tational routine must be acknowledged. It is assumed that the torso is electrically homogeneous. Although the actual presence of inhomogeneities has been shown to reduce the accuracy of the method," there is no scheme to incorporate them into the model. Their effects are virtually eliminated, however, when considering changes in dipole terms during the cardiac cycle of a single subject. Additionally, the geometric simplifications, although computationally critical, do introduce errors. That these errors are small in magnitude has been demonstrated in the studies referred to above reporting accurate inverse localization of experimental bipoles, Third, there is no independent method against which to compare computed dipole locations and moments. Thus, calculated coordinates at a given instant may be best compared to locations at other instants in the same subject. Comparisons between individuals are admittedly more speculative. Data presented above demonstrate that the normal cardiac repolarization source is fixed in location and orientation during much of the ST-T interval. This is in contrast to the wide and rapid swings in location determined during excitation (Fig. 2). Both observations are in direct correspondence with findings reported from isolated rabbit heart protocolai-s the equivalent QRS dipole moved an average distance of 0.42 inches, whereas the ST-T dipole moved only 0.08 inches during the initial two-thirds of ventricular recovery. Explanations for both the mobility of QRS dipoles and the stability of ST-T vectors may be based upon reported canine epicardial and intramural potential distribution.s" Three features are relevant to the present effort. First, multiple spatially separate and rapidly moving activation fronts were identified during much of the QRS. This would cause wide fluctuations in equivalent dipole locations not only due to physiologic effects but also to conceptual ones, resulting from significant extradipolar forces.iSecond, the epicardium was enveloped by a uniform, diffuse and stable positve field during much of the ST-T interval, associated with a undirectional transmural potential gradient. As has been discussed,14-22 this would yield a highly dipolar and stationary electrical field generated by an electromotive rim corresponding to the inactive basal cardiac circumference (Fig. 6 in reference 14). Third, during periods of overlap between excitation and recovery patterns, i.e., near the J. ELECTROCARDIOLOGY, VOL. 14, NO.4, 1981

395

J-point, epicardial patterns were complex and variable. Such inconstancy has been likewise observed in clinical body surface distributionsw and may underlie the instability of the computed heart vector during early ST-Tin these normal subjects. A second finding was the compactness of the grouping of dipole loci and directions during the ST-T interval. This was documented both by the small standard deviations in coordinates and angles presented in Fig. 4 and by the small summedsquared errors in patterns depicted in Fig. 5. Such closeness and the described stability of locations may aid the use of incompletely corrected orthogonallead systems. In addition, the narrow range of computed positions in this presumably homogeneous subject population, plus the correspondence of ranging parameters and isopotential distributions (Fig. 3), support the physiologic validity of the method. Results derived from the population with acute ischemic disease are similar to those of normals in some ways but different in others. The stability of dipole locations during the midportion of the ST-T interval (Figs. 6 and 8) would be expected from the anatomically stable boundary between ischemic and non-ischemic tissues. Movement during the ending of repolarization may be due to abnormally early recovery of ischemic tissue.sresulting in openings in the otherwise uniformly active epicardial shell. As in normals, these results also directly correspond with experimental data42 showing dipole location stability early and instability late in repolarization of isolated rabbit hearts after coronary ligation. A second finding was the greater dipole moment in subjects with anterior than with inferior lesions. Although the population studies were small, these data do relate to previous findings that 1) infarct size tends to be greater with anterior lesionss" and 2) dipole moment correlates directly with areas of nonperfused epicardial tissue.P Last, it was shown that dipole ranging permitted separation of electrocardiographically classified lesions by anatomic positions of computed heart vectors. Dipole loci in patients considered to have "inferior" lesions by the pattern of STsegment elevation in standard ECG records were located caudal, posterior and rightward of those computed from patients with "anterior" infarctions (Fig. 9). These differences correspond to the relative anatomic positions of anterior and inferior infarctions with the chest. 44 Similarly, di-

396

MIRVIS AND HOLBROOK

pole orientations were consistent with and endocardial-to-epicardial vector direction, as is generally assumed to occur in injured tissue. Thus, dipole ranging methods in man can provide physiologically relevant electrocardiologic information of potential clinical importance.

14.

15. 16.

REFERENCES 1. PARDEE, HE B: Electrocardiographic sign of coro2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

nary artery obstruction. Arch Intern Med 26:244, 1970 WILSON, F N, JOHNSTON, F D, ROSENBAUM, F F, ERLANGER, H, KOSSMANN, C E, HECHT, H H, COTRIM, N, MENDES DE OLIVERIA, R, SCAReI, RAND BARKER, P S: The precordial electrocardiogram. Am Heart J 27:19, 1944 MAROKO, P R, LIBBY, L J, COVELL, J W, SOBEL, B E, Ross, J AND BRAUNWALD, E: Precordial SoT segment elevation mapping: An atraumatic method for assessing alterations in the extent of myocardial ischemic injury. Am J Cardiol 22:223, 1972 AKIYAMA, T, HODGES, M, BIDDLE, T C, ZAWROTNY, BAND VANGELLOW, C: Measurement of ST segment elevation in acute myocardial infarction in man. Am J. Cardiol 36:155, 1975 . MULLER, J E, MAROKO, P R AND BRAl.i·NWALD, E: Precordial electrocardiographic mapping. A technique to assess the efficacy of interventions designed to limit infarct size. Circulation 57:1, 1978 GABOR, D AND NELSON, C V: Determination of the resultant dipole of the heart from measurements on the body surface. J Appl Physics 75:413, 1954 MIRVIS, D M: Body surface distribution of electrical potential during atrial depolarization and repolarization. Circulation 62:167, 1980 NELSON, C V, GASTONGUAY, P R, WILKINSON, A F AND VOUKYDIS, P C: A lead system for direction and magnitude of the heart vector. In Proceedings IXth International Vectorcardiography Symposium, I Hamby, Ed. North Holland Pub Co, Amsterdam, 1971, pp 85-97 NELSON, C V, HODGKIN, B C AND VOUKYDIS, P C: Determination of the locus of the heart vector from body surface measurements. J Electrocardiol8: 135, 1975 McFEE, R AND RUSH, S: Qualitative effects of thoracic resistivity variations on the interpretation of electrocardiograms. The low resistance surface layer. Am Heart J 76:48, 1968 OKADA, R H: Potentials produced by an eccentric current dipole in a finite-length circular conducting cylinder. IRE Trans Med Elect 7:14, 1956 BALL, G H AND HALL, D J: A clustering technique for summarizing multivariate data. Behav Sci 12:153, 1967 TACCARDI, B: Distribution of heart potentials on

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

the thoracic surface of normal human subjects. Circ Res 12:341, 1963 BRODY, D A, WARR, 0 S, WENNEMARK, J R, Cox, J W, KELLER, R W AND TERRY, F H: Studies of the equivalent cardiac generator behavior of isolated turtle hearts. Circ Res 29:512, 1971 TACCARDI, B: Distribution of heart potentials on dog's thoracic surface. Circ Res 11:862, 1962 HORAN, L G, FLOWERS, N C AND MILLER, C G: A rapid assay of dipolar and extradipolar content in the human electrocardiogram. J Electrocardiol 5:211,1974 TACCARDI, B: Body surface distribution of equipotential lines during atrial depolarization and ventricular repolarization. Circ Res 19:805, 1966 BRODY, D A, MIRVIS, D M, IDEKER, R D, Cox, J W, KELLER, F W, LARSEN, R A AND BANDURA, J P: Relative dipolar behavior of the equivalent T wave generator. Circ Res 40:263, 1977 EINTHOVEN, W, FAHR, G AND DE WAART, A: Uber die Richtung und die manifeste Grosse der Potential schwankugen im menschlichen Herzen und tiber den Einfluss der Herzloge auf die Form der Elektrokardiograms. Pflugers Arch Ges Physiol 150:275, 1913 BRODY, D A, Cox, J W, KELLER, F W AND WEN. NEMARK, J R: Dipole ranging in isolated rabbit hearts before and after right bundle branch block. Cardiovasc Res 8:37, 1974 IDEKER, R E, BANDURA, J P, LARSEN, R A, Cox, J W, KELLER, F W AND BRODY, D A: Localization of heart vectors produced by epicardial burns and ectopic stimuli. Circ Res 36:105, 1972 IDEKER, R E, BANDURA, J P, Cox, J W, KELLER, F W, MIRVIS, D M AND BRODY, D A: Path and significance of heart vector migration during QRS and ST-T complexes of ectopic beats in isolated perfused rabbit hearts. Circ Res 41:558, 1977 GASTONGUAY, P R AND NELSON, C V: Method for measuring dipole parameters of isolated hearts and muscles-Application to frog gastrocnemius. IEEE Trans Biomed Eng 15:289, 1968 NELSON, C V, ANGELAKOS, E T ANDGASTONGUAY, P R: Dipole moments of dog, monkey and lamb hearts. Circ Res 17:168, 1965 NELSON, C V, WAGGONER, W C ANDGASTONGUAY, P R: Dipole moments of isolated rabbit heart. Am J Physiol 208:250, 1965 DARKE, P G G AND HOLMES, J R: Studies on the equine cardiac electrical field. II. Integration of body surface potentials to derive resultant cardiac dipole moments. J Electrocardiol 2:235, 1969 NELSON, C V, HODGKIN, B C AND GASTONGUAY, P R: Dipole moment of the hearts of various species. Am Biomed Eng 3:308, 1975 HODGKIN, B C, NELSON, C V AND ANGELAKOS, E T: Magnitude, direction and location of the resultant dipole moment of the pig heart. J Electrocardiol 9:123, 1976

J. ELECTROCARDIOLOGY, VOL. 14, NO.4, 1981

DIPOLE RANGING IN MAN

29 . GESELOWlTZ, D G: Two theorems concerning the quadruple applicable to electrocardiography. IEEE Trans Biomed Eng 12:164, 1965

38.

30 . ARTHUR, R M, GESELOwrrz, D B, BRILLER, S A AND TROST, R F: The path of the electrical center of the human heart determined from surface electrocardiograms. J Electrocardiol 4:29 , 1971

39.

31. SCHUBERT, R W: An experimental study of the multipole series that represents the human electrocardiogram. IEEE Trans Biomed Eng 15:303, 1968 32 . TROST, R F, ARTHUR, R M, GESELOWITZ, D BAND BRILLER, S A: A dipole plus quadrupole lead system for human electrocardiography. J Electrocardiol 10:27,1977 33 . HORAN, L G, HAND, R C, FLOWERS, N C AND JOHNSON, J C: The multipolar content of the human electrocardiogram. Ann Biomed Eng 4:280, 1976 34 . BARBER, M R AND FISCHMAN, E J: A lead system recording total outward cardiac dipole strength. Brit Heart J 23:649, 1961 35 . ELLISON, R C, FISCHMAN, E J , MIETIINEN, 0 SAND HUGENHOLTZ, P G: Use of dipole moment in the assessment of left ventricular hypertrophy. Circulation 40:719, 1979 36 . HOLT, J H, BARNARD, A C L, LYNN, M S AND SVENDSEN, P: A study of the human heart as a multiple dipole electrical source. Circulation 40:687, 1969 37. BARNARD, A C L, MERRILL, A J, HOLT, J A AND KRAMER, J 0 : Progress in the evaluation of multipole dipole electrocardiography in a clinical environment. In Proceedings IXth International Vectorcardiography Symposium, I Hamby, ed.

J. ELECTROCARDIOLOGY, VOL. 14, NO.4, 1981

40 .

41.

42.

43.

44 .

397

North Holland Pub Co, Amsterdam, 1971, pp 404411 BRODY, D A: The inverse determination of simple generator configuration from equivalent dipole and multipole information. IEEE Trans Biomed Eng 15:106, 1968 SPACH, M S AND BARR, R C: Ventricular intramural and epicardial potential distributions during ventricular activation and repolarization in the intact dog. Circ Res 37:243, 1975 SPACH, M S, BARR, R C, BENSON, D W, WALSTON, A, WARREN, R B AND EDWARDS, S B: Body surface low-level potentials during ventricular repolarization with analysis of the ST segment. Variability in normal subjects. Circulation 59:822, 1979 DOWNAR, E, JANSE, M J AND DURRER, D: The effect of acute coronary artery occlusion on subepicardial transmembrane potentials in the intact porcine heart. Circulation 56:217, 1977 MmVIS, D M, KELLER, F W, IDEKER, R E, ZETrERGREN, D G AND DOWDIE, R F : Equivalent generator properties of acute ischemic lesions in the isolated rabbit heart. Cir Res 42:676, 1978 WARD, R M,IDEKER, R E, WAGNER, G S, ALONSO, D R, BISHOP, S P, BLOOR, C M, FALLON, J T, GoTrLIEB, G J, HACKEL, D B, PHILLIPS, H R, REIMER, K A, ROARK, S F, ROGERS, W J, RUTH, W K, SAVAGE, R M AND SELVESTER, R H: Myocardial infarcts in the lateral third of the left ventricle: Size and ECG recognition. Am J Cardiol 45:473, 1980 SAVAGE, R M, WAGNER, G , IDEKER, R E, PODOLSKY, S A AND HACKEL, D B: Correlation of postmortem anatomic findings with electrocardiographic changes in patients with myocardial infarction. Circulation 55:279, 1977