Francis W. Keller

Detection and localization of multiple epicardial electrical generators by a two-dipole ranging technique.

The ability of a numerical procedure to detect and to localize two experimentally induced, epicardial dipolar generators was tested in 24 isolated, perfused rabbit heart preparations, suspended in an electrolyte-filled spherical tank. Electrocardiograms were recorded from 32 electrodes on the surface of the test chamber before and after placement of each of two epicardial burns. The second lesion was located either 180 degrees, 90 degrees, or 45 degrees from the first. Signals were processed by iterative routines that computed the location of one or two independent dipoles that best reconstruced the observed surface potentials. The computed single dipole acounting for 99.68% of root mean sequare (RMS) surface potential recorded after the first burn was located 0.26 +/- 0.10 cm from the centroid of the lesion. Potentials recorded after the second lesions were fit with two dipoles that accounted for 99.36 +/- 1.51% of RMS surface potentials and that were located 0.42 +/- 0.26 cm and 0.57 +/- 0.49 cm from the centers of the corresponding burn. Seventy-one percent of computed dipoles were located within the visible perimeter of the burn. Thus, two simultaneously active dipolar sources can be detected and accurately localized by rigorous study of the generated electrical field.

Localization of heart vectors produced by epicardial burns and ectopic stimuli; validation of a dipole ranging method.

Location of the equivalent cardiac dipole has been estimated but not fully verified in several laboratories. To test the accuracy of such a procedure, injury vectors were produced in 14 isolated, perfused rabbit hearts by epicardial searing. Strongly dipolar excitation fronts were produced in 6 additional hearts by left ventricular pacing. Twenty computer-processed signals, derived from surface electrodes on a spherical electrolyte-filled tank containing the test preparation, were optimally fitted with a locatable cardiac dipole that accounted for over 99% of the root-mean-square surface potential. For the 14 burns (mean radius 5.0 mm), the S-T injury dipole was located 3.4 ± 0.7 (SD) mm from the burn center. For the 6 paced hearts, the dipole early in the ectopic beat was located 3.7 mm (range 2.6 to 4.6 mm) from the stimulating electrode. Phase inhomogeneities within the chamber appeared to have a small but predictable effect on dipole site determination. The study demonstrates that equivalent dipole location can be determined with acceptable accuracy from potential measurements of the external cardiac field.

Studies of the Equivalent Cardiac Generator Behavior of Isolated Turtle Hearts

Equivalent cardiac generator components were determined for a series of excised turtle hearts immersed in Ringers solution. Relatively large preparations were contained within a specially designed spherical chamber, and electric field potentials were derived from 20 evenly spaced electrodes on the chamber wall. A laboratory computer was used to acquire and store the 20 leads of signal data in digitized form in real time. An eccentric dipole was optimally fitted to the surface potentials for each 2 msec sample; the remaining voltages were used to determine a centric multipolar series through octapolar content. In addition to the purely quantitative parameters which were thus determined, sequential mapping of isopotential distribution over the spherical boundary gave valuable qualitative insights into the behavior of the equivalent generator throughout ventricular depolarization. This activity varied in complexity from predominantly dipolar to strongly nondipolar among different preparations. Peak quadripolar activity ranged from a low of 10 to a high of 60%; the corresponding figures for octapole content were 5−61%. The overall technique permits experimental exploration of several theoretical principles which have been advanced since 1954. Pilot studies on rabbit hearts indicate that the method will also be applicable to mammalian hearts.

Values and limitations of surface isopotential mapping techniques in the detection and localization of multiple discrete epicardial events

Summary The ability of surface isopotential mapping techniques to detect and to localize single and multiple discrete epicardial events was assessed using an isolated, perfused rabbit heart technique. Injury currents were generated by searing the epicardium at two sites separated by approximately 180°, 90° or 45° of rotation in three sets of eight preparations each. Electrocardiographic signals, recorded from 32 electrodes on the surface of the spherical test chamber, were processed by digital computer techniques to construct isopotential maps. Relative dipolarity of the surface pattern was quantitated by a numerical procedure and expressed as the percentage of root-mean-square (RMS) potential attributable to a single dipole source. After placing one burn, surface patterns demonstrated a single potential maximum spatially aligned with the lesion. In the 16 preparations with a second lesion 180° or 90° from the first, two discrete maxima appeared. The maximum related to the second lesion was located overlying the burn; that due to the first remained stationary but decreased in intensity by 67.86 ±23.78 μV (p 0.1). Thus, surface mapping techniques accurately depict single and dual generator locations unless the two sources are insufficiently separated. Maps depicting the differences in potential before and after the second lesion was induced did, however, demonstrate a maximum spatially aligned with the second burn in all subgroups. The roles of dipole strength, eccentricity and separation in determining surface patterns produced by two simultaneously active dipoles are further explored in an appended numerical simulation.

Path and significance of heart vector migration during QRS and ST-T complexes of ectopic beats in isolated perfused rabbit hearts.

Heart vector location was estimated for 23 isolated, perfused rabbit hearts during paced ectopic beats. Twenty computer-processed signals, derived from the surface electrodes of a spherical electrolyte-filled tank containing the hearts, were optimally fitted with a locatable cardiac dipole every millisecond of the QRS and every 3rd msec of the ST-T interval. During the QRS, the computed heart vector location of hearts subepicardially paced from the left ventricular free wall originated very close to the stimulating electrode, traversed the heart from left to right, and terminated in the right ventricle. Daring the first portion of repolarization for the hearts paced from the left ventricle, the position of the heart vector was almost stationary within the left ventricle, whereas after the peak of the T wave, heart vector location again moved from left to right. The first quarter of the QRS interval for hearts stimulated from the right ventricular free wall was nondipolar; during the remaining three-quarters of excitation, location of the heart vector moved from right to left, terminating in the left ventricle. Throughout the entire T wave of hearts paced from the right ventricle, the position of the heart vector remained almost motionless within the left ventricle. This study demonstrates the ability of heart vector location, by its rapid motion, graphically to portray passage of an ectopic beat across the heart and, by its slower motion within the central portion of the heart, to indicate the diffuse nature of the resulting ventricular recovery.

Equivalent generator properties of acute ischemic lesions in the isolated rabbit heart.

Equivalent generator properties of the electrical field produced by ischemic myocardium were studied in 25 isolated rabbit heart preparations. Electrocardiograms from 32 electrodes deployed about a spherical tank containing the isolated, perfused heart were recorded before and after suture ligation of the left anterior descending artery. The ligation was high (10 preparations) or low (10 preparations) along the interventricular septum. In a final five hearts, ligatures were placed sequentially in both positions. Signals were processed to quantitate the percentage of summed square (SSQ) potential attributable to a centric dipole (CD), to a four-element centric multipole series (CMS) and to a single moving dipole (SMD) during the S-T segment. Fifteen minutes after low ligation, 74 ± 12%, 98 ± 1%, and 96 ± 3% (mean ± 1 SD) of SSQ potential recorded 10 msec into the S-T segment could be accounted for by the CD, CMS, and SMD models, respectively. The computed SMD was determined to be fixed in location and orientation throughout the S-T segment but to increase in magnitude from early to late S-T. Dipole moment directly correlated with the area of the epicardial lesion (r = 0.82). Results were quantitatively similar after high ligation and after each of the sequential occlusions. Tank surface isopotential maps during the S-T segment uniformly demonstrated a single maximum that was spatially aligned with the ischemic lesion, and with an intensity proportional to both computed dipole moment and epicardial lesion size (r = 0.97). Thus, CMS and SMD cardiac models provide quantitatively accurate descriptions of the experimentally induced electrical fields.

Dipole, Quadripole, and Octapole Measurements in Isolated Beating Heart Preparations

An experimental technique is described for measuring the equivalent multipole model for isolated beating hearts. The dipole, quadripole, and octapole moments are measured from weighted sums of 20 electrode potentials on the surface of a 6.35-cm diameter sphere. An iterative procedure is used to eliminate the effects of source eccentricity for dipolar data.

Relative dipolar behavior of the equivalent T wave generator: quantitative comparison with ventricular excitation in the rabbit heart.

We studied the relative dipolar and nondipolar content of signal energy throughout ventricular excitation and recovery in 34 isolated, perfused rabbit hearts, suspended in an electrolyte-filled spherical chamber. Computer-processed signals were derived from 20 evenly spaced tank-surface electrodes, and a single, moving, equivalent cardiac dipole generator was optimally fitted to the recorded potentials for each 1-msec sampling interval. Superimposed, time-based plots of signal energy for the 34 preparations showed ventricular excitation to be strikingly more nondipolar than was recovery. In terms of the summed square ratio of nondipolar residual energies, overall nondipolarity of QRS exceeded that of ST-T by 41%. Furthermore, the maximum instantaneous ratio during QRS was considerably greater than during the ST-T. Evaluation of paired differences, comparing nondipolar behavior throughout QRS with all of ST-T, proved highly significant (P < .005). We also found that in contrast to the considerable mobility exhibited by the equivalent QRS dipole, the ST-T dipole locus remained nearly stationary during most of ventricular recovery. Presumably because repolarization is temporally and spatially a relatively diffuse process, it may generate electrical fields which are notably more dipolar than those caused by depolarization.

Equivalent cardiac generators: Two moving dipoles and moving dipole and quadripole

Two advanced models of equivalent cardiac generators were formulated and tested. The first model consists of two moving dipoles in a bounded spherical volume conductor. The second model consists of a simultaneous moving dipole and quadripole in an identical bounded spherical medium. The development of these models evolves from the derivation of the potential function for a dipole, a quadripole and an octapole in a homogeneous medium with a spherical dielectric boundary using a Taylor series expansion of a potential function due to two point current sources. The nonlinear models were tested in an inverse sense using ideal data and an existing nonlinear estimation scheme. Preliminary results indicate that such models may potentially be useful for characterization of the cardiac generator.

Experimental comparison of four inverse electrocardiographic constructs in the isolated rabbit heart.

Summary Complex models of the hearts electrical activity have been proposed as being superior to the fixed-location-dipole equivalent cardiac generator. To test and to compare the adequacy of four such proposals, i.e., a four element centric multipole series (CMS), a single moving dipole (SMD), two moving dipoles (TMD), and a moving dipole-quadripole pair (DPQP), the electrical fields generated by thirty isolated, perfused rabbit hearts placed in a spherical volume conductor were studied. Waveforms recorded from 32 surface electrodes were sampled 2500 times per second per channel, and generator parameters were computed using previously reported methods. A CMS accounted for 99.5±0.21% (mean±S.E.M.) of sum-squared surface potential recorded during ventricular depolarization. The dipole, quadripole, octapole, and hexadecapole terms fit 3.3 to 98.2%, 7.2 to 70.8%, 0.2 to 29.9% and 0.02 to 12.7% of observed potential, respectively. A SMD fit 90.9 ±0.25% of surface activity, whereas the TMD and DPQP constructs accounted for 98.4 ±0.21% and 99.2±0.21% of surface potential, respectively. The CMS accounted for significantly more (p