WO1996025094A1

WO1996025094A1 - Systems and methods for analyzing biopotential morphologies in body tissue

Info

Publication number: WO1996025094A1
Application number: PCT/US1996/002034
Authority: WO
Inventors: David K. Swanson; Dorin Panescu; James G. Whayne
Original assignee: Ep Technologies, Inc.
Priority date: 1995-02-17
Filing date: 1996-02-16
Publication date: 1996-08-22

Abstract

Systems and methods analyze bio-potential morphologies in body tissue. The systems and methods use a template of a bio-potential event of known cause in body tissue. The systems and methods compare this template to a sample of a bio-potential event externally triggered in body tissue. The systems and methods generate an output based upon this comparison. The systems and methods can be used to compare an event specific to the template of a cardiac event (T(i)) of known diagnosis to a sample of a placed cardiac event (P(i)). The comparison yields a matching coefficient (MCOEFF(i)) indicating how alike the input sample is to the input template. The matching coefficient can be used by the physician, for example, to aid in the location of sites that are potentially appropriate for ablation.

Description

SYSTEMS AND METHODS FOR ANALYZING BIOPOTENTIAL MORPHOLOGIES IN BODY TISSUE Field of the Invention The invention generally relates to systems and methods for acquiring, measuring, and analyzing biopotential morphologies in body tissue, and, in particularly, in heart tissue. Background of the Invention Normal sinus rhythm of the heart begins with the sinoatrial node (or "SA node") generating a depolarization wave front. The impulse causes adjacent myocardial tissue cells in the atria to depolarize, which in turn causes adjacent myocardial tissue cells to depolarize. The depolarization propagates across the atria, causing the atria to contract and empty blood from the atria into the ventricles. The impulse is next delivered via the atrioventricular node (or "AV node") and the bundle of HIS (or "HIS bundle") to myocardial tissue cells of the ventricles. The depolarization of these cells propagates across the ventricles, causing the ventricles to contract.

This conduction system results in the de- scribed, organized sequence of myocardial contrac¬ tion leading to a normal heartbeat.

Sometimes aberrant conductive pathways develop in heart tissue, which disrupt the normal path of depolarization events. For example, anatom- ical obstacles in the atria or ventricles can disrupt the normal propagation of electrical impuls¬ es. These anatomical obstacles (called "conduction blocks") can cause the electrical impulse to degen¬ erate into several circular wavelets that circulate about the obstacles. These wavelets, called "reen¬ try circuits," disrupt the normal activation of the atria or ventricles. As a further example, local¬ ized regions of ischemic myocardial tissue may propagate depolarization events slower than normal myocardial tissue. The ischemic region, also called a "slow conduction zone," creates errant, circular propagation patterns, called "circus motion." The circus motion also disrupts the normal depolariza¬ tion patterns, thereby disrupting the normal con- traction of heart tissue.

The aberrant conductive pathways create abnormal, irregular, and sometimes life-threatening heart rhythms, called arrhythmias. An arrhythmia can take place in the atria, for example, as in atrial tachycardia (AT) or atrial flutter (AF) . The arrhythmia can also take place in the ventricle, for example, as in ventricular tachycardia (VT) .

In treating arrhythmias, it is essential that the location of the sources of the aberrant pathways (call foci) be located. Once located, the tissue in the foci can be destroyed, or ablated, by heat, chemicals, or other means. Ablation can remove the aberrant conductive pathway, restoring normal myocardial contraction. Today, physicians examine the propagation of electrical impulses in heart tissue to locate aberrant conductive pathways. The techniques used to analyze these pathways, commonly called "mapping," identify regions in the heart tissue, called foci, which can be ablated to treat the arrhythmia. One form of conventional cardiac tissue mapping techniques uses multiple electrodes posi¬ tioned in contact with epicardial heart tissue to obtain multiple electrograms. The physician stimu- lates myocardial tissue by introducing pacing sig¬ nals and visually observes the morphologies of the electrograms recorded during pacing, which this Specification will refer to as "paced electrograms." The physician visually compares the patterns of paced electrograms to those previously recorded during an arrhythmia episode to locate tissue regions appropriate for ablation. These conven¬ tional mapping techniques require invasive open heart surgical techniques to position the electrodes on the epicardial surface of the heart.

Conventional epicardial electrogram pro¬ cessing techniques used for detecting local electri¬ cal events in heart tissue are often unable to interpret electrograms with multiple morphologies. Such electrograms are encountered, for example, when mapping a heart undergoing ventricular tachycardia (VT) . For this and other reasons, consistently high correct foci identification rates (CIR) cannot be achieved with current multi-electrode mapping tech- nologies.

Another form of conventional cardiac tissue mapping technique, called pace mapping, uses a roving electrode in a heart chamber for pacing the heart at various endocardial locations. In searching for the VT foci, the physician must visually compare all paced electrocardiograms (recorded by twelve lead body surface electrocardiograms (ECG's)) to those previously recorded during an induced VT. The physician must constantly relocate the roving electrode to a new location to systematically map the endocardium.

These techniques are complicated and time consuming. They require repeated manipulation and movement of the pacing electrodes. At the same time, they require the physician to visually assimi¬ late and interpret the electrocardiograms.

Furthermore, artifacts caused by the pacing signals can distort the electrocardiograms. The pacing artifacts can mask the beginning of the Q- wave in the electrocardiogram. In body surface mapping, the morphology of the pacing artifact visually differs from the morphology of the electro¬ cardiogram. A trained physician is therefore able to visually differentiate between a pacing artifact and the electrocardiogram morphology. This is not always the case in endocardial or epicardial map¬ ping, in which there can be a very close similarity between the morphology of the pacing artifact and the bipolar electrogram morphology. Under the best conditions, the pacing artifact and electrogram complex are separated in time, and therefore can be distinguished from one another by a trained physi¬ cian. Under other conditions, however, the presence of the pacing artifact can sometimes mask the entire bipolar electrogram. In addition, its likeness to the bipolar electrogram often makes it difficult or impossible for even a trained physician to detect the beginning of depolarization with accuracy.

There thus remains a real need for cardiac mapping and ablation systems and procedures that simplify the analysis of electrograms and the use of electrograms to locate appropriate arrhythmogenic foci. aιιτnm»rv of the Invention A principal objective of the invention is to provide improved systems and methods to examine body tissue morphologies quickly and accurately.

The invention provides systems and methods for analyzing biopotential morphologies in body tissue. The systems and methods use a template of a biopotential event of known cause in body tissue. The systems and methods compare this template to a sample of a biopotential event externally triggered in body tissue. The systems and methods generate an output based upon the comparison.

In a preferred embodiment, the systems and methods analyze biopotentials in myocardial tissue. In this embodiment, the template comprises a tem¬ plate of a cardiac event of known diagnosis. In a preferred embodiment, the selected event comprises an arrhythmia that the physician seeks to treat, for example, ventricular tachycardia (VT) , or atrial tachycardia (AT) , or atrial fibrillation (AF) . The systems and methods compare this event-specific template to a sample of a paced cardiac event. In a preferred embodiment, entrainment pacing is used. Pace mapping can also be implemented. The comparison yields an output.

In a preferred embodiment, the output comprises a matching coefficient indicating how alike the input sample is to the input template. The matching coefficient can be used by the physi¬ cian, for example, to aid in the location of sites that are potentially appropriate for ablation. A matching coefficient that indicates close similarity between the sample and the template suggests that the pacing site lies close to a region appropriate for ablation to treat the arrhythmia.

The systems and methods employ various alternative methods for comparing the template to input, paced sample. Various alternative embodi¬ ments employ matched filtering, cross correlation, and norm of the difference techniques. In another alternative embodiment, the input sample is compared to the input template by using the input template to create a matched filtered sample and by analyzing the symmetry of the matched filtered sample.

The systems and methods lend themselves to the analysis of both electrograms and electrocardio- grams. In addition to processing or analyzing signals derived from cardiac-related events, the systems and methods can be used to process or analyze the biopotential morphologies in other body tissue, for example, in electroencephalograms, electrogastrograms, and electromyograms.

Other features and advantages of the inven¬ tions are set forth in the following Description and Drawings, as well as in the appended Claims. Brief Description of the Drawings Fig. 1A is a diagrammatic view of a system, which embodies the features of the invention, for accessing a targeted tissue region in the body for diagnostic or therapeutic purposes;

Fig. IB is a diagrammatic view of the system shown in Fig. 1A, with the inclusion of a roving pacing probe and additional features to aid the physician in conducting diagnosis and therapeu¬ tic techniques according to the invention;

Fig. 2 is an enlarged perspective view of a multiple-electrode structure used in association with the system shown in Fig. 1;

Fig. 3 is an enlarged view of an ablation probe usable in association with the system shown in Figs. 1A and IB; Fig. 4A is a diagrammatic view of the process controller shown in Figs. 1A and IB, which locates by electrogram matching a site appropriate for ablation;

Fig. 4B is a schematic view of a slow conduction zone in myocardial tissue and the circu¬ lar propagation patterns (called circus motion) it creates;

Fig. 5 is a flow chart showing a pattern matching technique that the process controller shown in Fig. 4A can employ for matching electrograms according to the invention;

Figs. 6A to 6E are representative electro¬ gram morphologies processed in accordance with the pattern matching technique shown in Fig. 5; Figs. 7A and 7B are, respectively, a flow chart and illustrative wave shape showing a symmetry matching technique that the process controller shown in Fig. 4A can employ for matching electrograms according to the invention; Figs. 8A to 8C are representative electro¬ gram morphologies processed in accordance with the symmetry matching technique shown in Fig. 7A;

Fig. 9 is a flow chart showing a matched filtering technique that the process controller shown in Fig. 4A can employ for matching electrograms according to the invention;

Fig. 10 is a flow chart showing a cross correlation coefficient technique that the process controller shown in Fig. 4A can employ for matching electrograms according to the invention;

Figs. 11A and 11B are representative elec¬ trogram morphologies processed in accordance with the cross correlation coefficient technique shown in Fig. 10; Fig. 12 is a flow chart showing a norm of the difference technique that the process controller shown in Fig. 4A can employ for matching electrograms according to the invention;

Figs. 13A and 13B are representative elec- trogram morphologies processed in accordance with the norm of the difference technique shown in Fig. 12;

Fig. 14A is a flow chart showing a filter¬ ing technique that the process controller shown in Fig. 4A can employ for removing pacing artifacts according to the invention;

Fig. 14B is a diagram showing the implemen¬ tation of the filtering technique shown in Fig. 14A as a median filter; Fig. 14C is a diagram showing the implemen¬ tation of the filtering technique shown in Fig. 14A as a non-median filter;

Fig. 15A to D; 16A to D; 17A to D are representative electrogram morphologies showing the effect of the sort position selection criteria in removing the pacing artifact employing the technique shown in Fig. 14;

Figs. 18A to C are representative elec¬ trogram morphologies showing the effect of the sample window size in removing the pacing artifact employing the technique shown in Fig. 14;

Fig. 19 is a diagrammatic view of an adap¬ tive filtering technique that the process controller shown in Fig. 4A can employ for removing pacing artifacts according to the invention;

Figs. 20A and 20B are representative elec¬ trogram morphologies processed by the adaptive filtering technique shown in Fig. 19 to remove a pacing artifact; Fig. 21 is a diagrammatic flow chart showing the operation of the time-sequential mode of the process controller shown in Fig. 1 in creating an electrogram composite over different time inter¬ vals; Figs. 22A to 22D show representative individual electrograms taken at different time intervals during the time-sequential mode shown in Fig. 21, before time-alignment;

Figs. 23A to 23D show the representative individual electrograms shown in Figs. 22A to 22D, after time-alignment, to form the electrogram composite;

Fig. 24 shows a pace electrogram with three pacing artifacts, before adaptive filtering; Figs. 25A to C show the artifact signals of the three pacing artifacts shown in Fig. 24, manual¬ ly selected by the physician, before alignment and truncation;

Figs. 26A to C show the artifact signals of the three pacing artifacts shown in Fig. 25 A to C, after alignment and truncation;

Fig. 27 shows the artifact template that results from averaging the three artifact signals shown in Figs. 26A to C; and Figs. 28A and B show the alignment of the first pacing artifact (in Fig. 28A) with the arti¬ fact template (in Fig. 28B) .

The invention may be embodied in several forms without departing from its spirit or essential characteristics. The scope of the invention is defined in the appended claims, rather than in the specific description preceding them. All embodi¬ ments that fall within the meaning and range of equivalency of the claims are therefore intended to be embraced by the claims. Description of the Preferred Embodiments

Fig. 1A shows the components of a system 10 for analyzing body tissue biopotential morphologies for diagnostic or therapeutic purposes. The illus- trated embodiment shows the system 10 being used to examine the depolarization of heart tissue that is subject to an arrhythmia. In this embodiment, the system 10 serves to locate an arrhythmogenic focus for removal by ablation. The invention is well suited for use in conducting electrical therapy of the heart.

Still, it should be appreciated that the invention is applicable for use in other regions of the body where tissue biopotential morphologies can be ascertained by analyzing electrical events in the tissue. For example, the various aspects of the invention have application in procedures for analyz¬ ing brain or neurologic tissue. Fig. 1A shows the system 10 analyzing endocardial electrical events, using catheter-based, vascular access techniques. Still, many aspects of the invention can be used in association with techniques that do not require any intrusion into the body, like surface electrocardiograms or electroencephalograms. Many of the aspects of the invention also can be used with invasive surgical techniques, like in open chest or open heart sur¬ gery, or during brain surgery.

In particular, Fig. 1A shows the system 10 analyzing electrical events within a selected region 12 inside a human heart. Figs. 1A and IB generally show the system 10 deployed in the left ventricle of the heart. Of course, the system 10 can be deployed in other .regions of the heart, too. It should also be noted that the heart shown in the Fig. 1 is not anatomically accurate. Figs. 1A and IB show the heart in diagrammatic form to demonstrate the features of the invention.

The system 10 includes a mapping probe 14 and an ablation probe 16. In Fig. 1A, each is sepa¬ rately introduced into the selected heart region 12 through a vein or artery (typically the femoral vein or artery) through suitable percutaneous access. Alternatively, the mapping probe 14 and ablation probe 16 can be assembled in an integrated structure for simultaneous introduction and deployment in the heart region 12.

Further details of the deployment and structures of the probes 14 and 16 are set forth in pending U.S. Patent Application Serial No. 08/033,641, filed March 16, 1993, entitled "Systems and Methods Using Guide Sheaths for Introducing, Deploying, and Stabilizing Cardiac Mapping and Ablation Probes." The mapping probe 14 has a flexible cathe¬ ter body 18. The distal end of the catheter body 18 carries a three dimensional multiple-electrode structure 20. In the illustrated embodiment, the structure 20 takes the form of a basket defining an open interior space 22 (see Fig. 2) . It should be appreciated that other three dimensional structures could be used.

As Fig. 2 shows, the illustrated basket structure 20 comprises a base member 26 and an end cap 28. Generally flexible splines 30 extend in a circumferentially spaced relationship between the base member 26 and the end cap 28.

The splines 30 are preferably made of a resilient, biologically inert material, like Nitinol metal or silicone rubber. The splines 30 are con- nected between the base member 26 and the end cap 28 in a resilient, pretensed, radially expanded con¬ dition, to bend and conform to the endocardial tissue surface they contact. In the illustrated em- bodiment (see Fig. 2) , eight splines 30 form the basket structure 20. Additional or fewer splines 30 could be used.

The splines 30 carry an array of electrodes 24. In the illustrated embodiment, each spline 30 carries eight electrodes 24. Of course, additional or fewer electrodes 24 can be used.

A slidable sheath 19 is movable along the axis of the catheter body 18 (shown by arrows in Fig. 2) . Moving the sheath 19 forward causes it to move over the basket structure 20, collapsing it into a compact, low profile condition for introducing into the heart region 12. Moving the sheath 19 rearward frees the basket structure 20, allowing it to spring open and assume the pretensed, radially expanded position shown in Fig. 2. The electrodes are urged into contact against the surrounding heart tissue.

Further details of the basket structure are disclosed in pending U.S. Patent Application Serial No. 08/206,414, filed March 4, 1994, entitled "Multiple Electrode Support Structures."

In use, the electrodes 24 sense electrical events in myocardial tissue for the creation of electrograms. The electrodes 24 are electrically coupled to a process controller 32 (see Fig. 1A) . A signal wire (not shown) is electrically coupled to each electrode 24. The wires extend through the body 18 of the probe 14 into a handle 21, in which they are coupled to an external multiple pin connec- tor 23. The connector 23 electrically couples the electrodes to the process controller 32.

Alternatively, multiple electrode structures can be located epicardially using a set of catheters individually introduced through the coronary vasculature (e.g., retrograde through the aorta or coronary sinus) , as disclosed in PCT/US94/01055 entitled "Multiple Intravascular Sensing Devices for Electrical Activity."

The ablation probe 16 (see Fig. 3) includes a flexible catheter body 34 that carries one or more ablation electrodes 36. For the sake of illustra¬ tion, Fig. 3 shows a single ablation electrode 36 carried at the distal tip of the catheter body 34. Of course, other configurations employing multiple ablation electrodes are possible, as described in pending U.S. Patent Application Serial No. 08/287,310, filed August 8, 1994, entitled "Systems and Methods for Ablating Heart Tissue Using Multiple Electrode Elements." A handle 38 is attached to the proximal end of the catheter body 34. The handle 38 and catheter body 34 carry a steering mechanism 40 for selec¬ tively bending or flexing the catheter body 34 along its length, as the arrows in Fig. 3 show. The steering mechanism 40 can vary. For example, the steering mechanism can be as shown in U.S. Patent 5,254,088, which is incorporated herein by reference.

A wire (not shown) electrically connected to the ablation electrode 36 extends through the catheter body 34 into the handle 38, where it is electrically coupled to an external connector 45. The connector 45 connects the electrode 36 to a generator 46 of ablation energy. The type of energy used for ablation can vary. Typically, the genera- tor 46 supplies electromagnetic radio frequency energy, which the electrode 36 emits into tissue. A radio frequency generator Model EPT-1000, available from EP Technologies, Inc., Sunnyvale, California, can be used for this purpose.

In use, the physician places the ablation electrode 36 in contact with heart tissue at the site identified for ablation. The ablation electrode emits ablating energy to heat and thermally destroy the contacted tissue.

According to the features of the invention, the process controller 32 employs electrogram matching to automatically locate for the physician the site or sites potentially appropriate for abla- tion.

I. Electrogram Matching

The process controller 32 is operable to sense electrical events in heart tissue and to process and analyze these events to achieve the objectives of the invention. The process controller 32 is also selectively operable to induce electrical events by transmitting pacing signals into heart tissue.

More particularly, the process controller 32 is electrically coupled by a bus 47 to a pacing module 48, which paces the heart sequentially through individual or pairs of electrodes to induce depolar¬ ization. Details of the process controller 32 and pacing module 48 are described in copending U.S. Patent Application Serial No. 08/188,316, filed January 28, 1994, and entitled "Systems and Methods for Deriving Electrical Characteristics of Cardiac

Tissue for Output in Iso-Characteristic Displays."

The process controller 32 is also electrically coupled by a bus 49 to a signal processing module 50. The processing module 50 processes cardiac signals into electrograms. A Model TMS 320C31 processor available from Spectrum Signal Processing, Inc. can be used for this purpose.

The process controller 32 is further electrical- ly coupled by a bus 51 to a host processor 52, which processes the input from the electrogram processing module 50 in accordance with the invention to locate arrhythmogenic foci. The host processor 32 can comprise a 486-type microprocessor. According to the invention, the process con¬ troller 32 operates in two functional modes, called the sampling mode and the matching mode.

In the sampling mode, the physician deploys the basket structure 20 in the desired heart region 12. To assure adequate contact is made in the desired region 12, the physician may have to collapse the basket structure 20, rotate it, and then free the basket structure 20. The degree of contact can be sensed by the process controller 32 in various ways. For example, the process controller 32 can condition the pacing module 48 to emit pacing signals through a selected electrode 24 or pair of electrodes 24. The process controller 32 conditions the electrodes 24 and processing module 50 to detect electrograms sensed by a desired number of the electrodes 24. The processing module can also ascertain the desired degree of contact by measuring tissue impedance, as described in co-pending patent application Serial No. 08/221,347, filed March 31, 1994, and entitled "Systems and Methods for Positioning Multiple Electrode Structures in Electrical Contact with the Myocardium. "

Once the basket structure 20 is properly posi¬ tioned, the process controller 32 conditions the electrodes 24 and signal processing module 50 to - 16 -

record electrograms during a selected cardiac event having a known diagnosis. In the sampling mode, the process controller 32 typically must condition the pacing module 48 to pace the heart until the desired cardiac event is induced. Of course, if the patient spontaneously experiences the cardiac event while the structure 20 is positioned, then paced-induction is not required. The processor controller 32 saves these electrograms in the host processor 52. The process controller 32 creates templates of selected electrogram morphologies by any conventional method, e.g., by having the physician manually select representative electrogram morphologies. At the end of the sampling mode, the process controller 32 typically must condition the pacing module 48 to pace terminate the cardiac event, or the physician may apply a shock to restore normal sinus rhythm.

The matching mode is conducted without altering the position of the multiple electrode structure 20 in the heart region 12, so that the electrodes 24 occupy the same position during the matching mode as they did during the sampling mode.

In the matching mode, the process controller 32 conditions the pacing module 48 to pace the heart in a prescribed manner without inducing the cardiac event of interest, while conditioning the signal processing module 50 to record the resulting electrograms. The process controller 32 operates the host processor 52 to compare the resulting paced electrogram morphologies to the electrogram morphol¬ ogy templates collected during the sampling mode. Based upon this comparison, the host processor 52 generates an output that identifies the location of the electrode or electrodes 24 on the structure 20 that are close to a potential ablation site. A. The Sampling Mode

As before generally described, the process con¬ troller 32 operates in the sampling mode while the heart is experiencing a selected cardiac event of known diagnosis and the basket structure 20 is retained in a fixed location in the region 12. In the illustrated and preferred embodiment, the selected event comprises an arrhythmia that the physician seeks to treat, for example, ventricular tachycardia (VT) , or atrial tachycardia (AT) , or atrial fibrillation (AF) .

As Fig. 4A shows, during the sampling mode, the signal processing module 50 processes the electro¬ gram morphologies obtained from each electrode during the known cardiac event (designated for the purpose of illustration as El to E3 in Fig. 4A) . The electrograms may be recorded unipolar (between an electrode 24 and a reference electrode, not shown) or bipolar (between electrodes 24 on the structure 20) .

The host processor 52 creates a digital, event- specific template for the morphology sensed at each electrode (designated for the purpose of illustra¬ tion as TI to T3 in Fig. 4A) . The event-specific templates TI to T3 for each electrode El to E3 can be based upon electrogram morphology from one heart beat or a specified number of heart beats. The event-specific template TI to T3 for each electrode El to E3 can be created by, for example, having the physician manually select representative electrogram morphologies.

If the arrhythmia event is not polymorphic, the template preferably comprises one heart beat and is updated beat by beat. Also preferably, though not essential, the starting point of the template should coincide with the beginning of the depolarization and extend one beat from that point. However, if the arrhythmia event under study is polymorphic, it may be necessary to extend the template over several beats. For example, in bigeminy cases, the template should preferably extend over two beats.

The host processor 52 retains the set of event- specific templates TI to T3 in memory. The processor 52 can, for an individual patient, retain sets of event-specific templates for different cardiac events. For example, a patient may undergo different VT episodes, each with a different morphology. The processor 52 can store templates for each VT episode for analysis according to the invention. The templates can be downloaded to external disk memory for off-line matching at a subsequent time, as will be described later. Templates can also be generated based upon mathematical modeling or empirical data and stored for later matching for diagnostic purposes.

B. The Matching Mode

In the matching mode, the process controller 32 operates the pacing module 48 to apply pacing signals sequentially to each of the individual elec- trodes. The pacing electrode is designated Ep in Fig. 4A.

The pacing signal induces depolarization, emanating at the location of the pacing electrode Ep. The process controller 32 operates the signal processing module 50 to process the resulting paced electrogram morphologies sensed at each electrode (again designated El to E3 for the purpose of illustration in Fig. 4A) during pacing by the selected individual electrode Ep. The processed paced electrograms are designated Pi to P3 in Fig. 4A .

The paced morphology PI to P3 at each electrode can be from one heart beat or a specified number of heart beats, provided that the length of the mor- phologies PI to P3 is not shorter than the length of the event-specific templates TI to T3 for the same electrodes El to E3 obtained during the sampling mode.

Different conventional pacing techniques can be used to obtain the paced morphologies PI to P3. For example, conventional pace mapping can be used, during which the pace rate is near the arrhythmia rate, but arrhythmia is not induced.

For reasons that will be explained later, conventional entrainment or reset pacing is the pre¬ ferred technique. During entrainment pacing, the pacing rate is slightly higher than and the period slightly lower than that observed during the ar¬ rhythmia event, thereby increasing the rate of the induced arrhythmia event. Further details of entrainment pacing are found in Almendral et al., "Entrainment of Ventricular Tachycardia: Explanation for Surface Electrocardiographic Phenomena by Analysis of Electrograms Recorded Within the Tachy- cardia Circuit," Circulation, vol. 77, No. 3, March 1988, pages 569 to 580, which is incorporated herein by reference.

Regardless of the particular pacing technique used, the pacing stimulus may be monophasic, biphasic, or triphasic.

In the matching mode, while pacing at an indi¬ vidual one of the electrodes Ep, the host processor 52 compares the paced morphology PI to P3 obtained at each electrode El to E3 to the stored event- specific template TI to T3 for the same electrode El to E3. The comparisons (which are designated CI to C3 in Fig. 4A) can be performed by using matched filtering or correlation functions, as will be described later. Alternatively, the paced morphologies PI to P3 can be retained in memory or downloaded to external disk memory for matching at a later time. To accommodate off-line processing, the host processor 52 preferably includes an input module 72 for uploading pregenerated templates and/or paced morphologies recorded at an earlier time. The input module 72 allows templates and paced morphologies to be matched off-line by the host processor 52, without requiring the real time presence of the patient. Alternatively, recorded paced morphologies can be matched in real time using templates generated earlier. The pregenerated templates can represent "typical" biopotential events based upon either real, empirical data, or mathematical models for diagnostic purposes, or reflect earlier biopotential events recorded for the same patient or for a patient having the same or similar prognosis.

For each pacing electrode Ep(j), the host processor 52 generates a matching coefficient M_C0EF -_j for each electrode E(i) from the comparison C(i) of the pacing morphology P(i) to the template morpholo¬ gy T(i) for the same electrode E(i) . Preferably, both j and i = 1 to n, where n is the total number of electrodes on the three dimensional structure (which, for the purpose of illustration in Fig. 4A, is 3) .

The value of the matching coefficient M_C0EF(j) is indicative for that electrode E(i) how alike the pacing morphology P(i) is to the event-specific tem- plate T(i) for that electrode E(i) . The value of M_j^_gp^_j for each electrode E(i) varies as the location of the pacing electrode Ep(j) changes. Generally speaking, the value of the matching coefficient _{' OEfli )} for a given electrode E(i) increases in rela- tion to the closeness of the pacing electrode Ep(j) to the arrhythmogenic foci. In the illustrated and preferred embodiment (as Fig. 4A shows) , while pacing at an individual one of the electrodes Ep(j), the host processor 52 generates from the matching coefficients M_C0EF(j) for each electrode E(i) an overall matching factor V^^^^. for the pacing elec¬ trode Ep(j) . The value of the overall matching factor M_pACE(J._j for the pacing electrode Ep(j) is indicative of how alike the overall propagation pattern observed during pacing at the electrode Ep(j) is to the overall propagation pattern recorded on the associated event-specific templates.

The process controller 32 operates the pacing module 48 to apply a pacing signal sequentially to each electrode Ep(j) and processes and compares the resulting electrogram morphologies at each electrode E(i) (including Ep(j)) to the event-specific tem¬ plates, obtaining the matching coefficients M_C0EF(i) for each electrode E(i) and an overall matching factor M_pACE(J) for the pacing electrode Ep(j) , and so on, until every electrode E(i) serves as a pacing electrode Ep(j) .

^M _PACE(j) ^{for eac1} pacing electrode can be derived from associated matching coefficients M_C0EF(j) in various ways.

For example, various conventional averaging techniques can be used. For example, M_pACE(j) can be computed as a first order average (arithmetic mean) ^{of M}co_EF(i> ^as follows: ΣM M ^{co i)}

PACE _n

where i = 1 to n; or as a weighted arithmetic mean, as follows:

^W®^MCOEI{,]

where i = 1 to n; Σ W(i) = l. If W(i) = l/n, for each i, then the arithmetic mean is obtained. Generally speaking, the value of the overall matching factor M_pACE(jj increases in relation to the proximity of the particular pacing electrode Ep(j) to a potential ablation site.

By way of overall explanation, for VT, the site appropriate for ablation typically constitutes a slow conduction zone, designated SCZ in Fig. 4B. Depolarization wave fronts (designated DWF in Fig. 4B) entering the slow conduction zone SCZ (at site A in Fig. 4B) break into errant, circular propaga- tion patterns (designated B and C in Fig. 4B) , called "circus motion." The circus motions disrupt the normal depolarization patterns, thereby disrupt¬ ing the normal contraction of heart tissue to cause the cardiac event. The event-specific templates T(i) record these disrupted depolarization patterns. When a pacing signal is applied to a slow conduction zone, the pacing signal gets caught in the same circus motion (i.e., paths B and C in Fig. 4B) that triggers the targeted cardiac event. A large proportion of the associated pacing morphologies P(i) at the sensing electrodes E(i) will therefore match the associated event-specific templates P(i) recorded during the targeted cardiac event. This leads to a greater number of larger matching coefficients M_C0EF(i) and thus to a larger overall matching factor M_pACE(J).

However, when a pacing signal is applied outside a slow conduction zone, the pacing signal does not get caught in the same circus motion. It propagates free of circus motion to induce a significantly different propagation pattern than the one recorded in the templates T(i) . A large proportion of the pacing morphologies P(i) at the sensing electrodes E(i) therefore do not match the event-specific tem¬ plates T(i) . This leads to a smaller number of larger matching coefficients M^^._j and thus to a smaller overall matching factor M_{PACE( )}. This is why the overall matching factor M_pACE(j;. becomes larger the closer the pacing electrode Ep(j) is to the slow conduction zone, which is the poten¬ tial ablation site. The difference in propagation patterns between pacing inside and outside a slow conduction zone is particularly pronounced during entrainment pacing. For this reason, entrainment pacing is preferred.

Ablating tissue in or close to the slow conduc¬ tion zone prevents subsequent depolarization. The destroyed tissue is thereby "closed" as a possible path of propagation. Depolarization events bypass the ablated region and no longer become caught in circus motion. In this way, ablation can restore normal heart function. The matching of pacing morphologies P(i) to tem¬ plate morphologies T(i) to create the matching coefficient M_C0EF(i) and the overall matching factor M_pACE(i) can be accomplished in various ways. According to the invention, the host processor 52 can employ pattern matching; symmetry matching; matched filter- ing; cross correlation; or norm of the difference techniques. The following provides an overview of each of these techniques.

1. Pattern Matching Fig. 5 diagrammatically shows a pattern matching technique that embodies features of the invention.

The pattern matching technique matched filters the template T(i) for each electrode E(i) using the same template flipped left to right with respect to time, Tflip(i) , as coefficients of the matched filter. Fig. 6B shows a representative template T(i) for a given electrode E(i) . Fig. 6C shows Tflip(i), which is the template T(i) (shown in Fig. 6B) flipped right to left. Fig. 6E shows a matched fil- tered output MT(i) , which had T(i) (Fig. 6B) as input and Tflip(i) (Fig. 6C) for the same electrode E(i) as coefficients of the matched filter. As Fig. 6E shows, the matched filtered output MT(i) is, for the electrode E(i), a sequence of alternating maxi- mums and minimums, with their values marking a first pattern employed by this technique.

The pattern matching technique also matched filters the paced electrogram P(i) for each elec¬ trode E(i) using an identical matched filter as the one described above. Fig. 6A shows a representative paced electrogram P(i) for the given electrode E(i) . Fig. 6D shows the matched filtered output MP(i), using Tflip(i) shown in Fig. 6C as the matched filtered coefficients. Like MT(i), the matched filtered output MP(i) is, for each electrode E(i), a sequence of alternating maximums and minimums, which are used to construct a second pattern.

The pattern matching technique detects the maxi¬ mums and minimums for the matched filtered template outputs MT(i) and those of MP(i). The pattern matching technique places the maximums and minimums in two odd-length, L-sized model vectors, with the largest excursions at positions

where L is the total number of local extremes of MT(i) and MP(i) . The pattern matching technique computes the norm of the difference between the MP- pattern and the corresponding MT-pattern shifted by an amount, P, that varies from -K to K, where K < L/2. The maximum number of comparisons for n elec- trodes will be n comparisons for each pacing elec¬ trode. Alternatively, one can shift the MP-patterns as just described, keeping the corresponding MT- patterns fixed. The largest excursions are placed in the centers of the template and paced vectors. For example, assuming:

MT(i)={mt(i,l) ,mt(i,2) ,mt(i,3) }, and MP(i)={mp(i,l+p) ,mp(i,2+p) ,mp(i,3+p) }, then norm_(i#p) = |MT(i) - MP(i)| or

where r = 1 to 3 and p = -K to K.

The minimum of the above is used as the matching coefficient for the sequence of norms (Mco_Eci₎) ' i.e. , :

^MCOEH ^=nUn "^0rm^p)

for p = -K to K. The minimum norms of the electrodes are averaged by an appropriate weighted average algorithm (as above discussed) . This yields the overall matching factor M_pACε(j) for each pacing electrode Ep(j), i.e.,

•

_M ^IYtPACE()_^ΣΛ ι) n

2. symmetry Matching Fig. 7A shows a symmetry matching technique that embodies features of the invention.

The symmetry matching technique matched filters the paced electrogram P(i) for each electrode E(i) using Tflip(i) as coefficients of the matched fil- ter. Each filter output is tested with respect to the largest excursion or extreme from baseline (EXC_MAX) , the sign (positive or negative) of EXC^_X , and a symmetry index (SYM) , where:

except when EXC_MAχ < 0, then SYM = 1; where N equals the number of local extremes to the left of EXC^ (which is also equal to the number of local extremes situated to the right of EXC_MAχ (see Fig. 7B) . The technique first determines whether EXC_MAχ > 0 (that is, whether it is positive) . If EXC_HAχ is not positive (i.e, SYM = 1.0), the technique deems that a poor match has occurred on this criteria alone. If ^EXC _MAX ^^s positive, the technique goes on to compute the symmetry index SYM and compares SYM to a symmetry threshold (SYM_THRESH) . If SYM < SYM._HRESH , the technique deems that a good match has occurred. In the preferred embodiment, SYM_THRESH = 0.2 (for perfect symmetry, SYM = 0.0).

Similar electrograms will create a matched filtered output having a positive largest excursion. As the degree of similarity between the two electrograms increases, the matched filtered output will become increasingly more symmetric about this positive absolute maximum. The scoring factor is created for each electrogram comparison, where the scoring factor M-a^_j) = 1 - SYM. The scoring factors based upon SYM are converted to an overall matching factor M_{pACE( )} for each pacing electrode Ep(j), as previously described. The pacing electrode Ep(j) creating the highest overall matching factor is designated to be close to a potential ablation site.

For example, Fig. 6E shows the matched filtered output MT(i) of the template electrogram of Fig. 6B and its left-to-right flipped counterpart of Fig. 6C. The electrogram of Fig. 6B is, in effect, matched filtered against itself, and the symmetry matching technique detects this. Fig. 6E shows a largest excursion that is positive and an output that is perfectly symmetric about the positive absolute maximum. A perfect scoring factor M_{C0EF )} of 1.0 would be assigned.

Refer now to Fig. 6D, which is the matched filtered output MP(i) of the electrogram of Fig. 6A and the flipped template in Fig. 6C. These are different, yet similar electrograms. The symmetry matching technique detects this close similarity. Fig. 6D shows a positive largest excursion, and the output is relatively symmetric about this positive absolute maximum. A good scoring factor M_C0EF(j) of, for example, 0.9 would be assigned. Refer now to Fig. 8C, which is the matched fil¬ tered output of the electrogram of Fig. 6A using the flipped template shown in Fig. 8B as coefficient of the matched filter. It can be seen that the elec- trogram shown in Fig. 8A has a morphology quite different than that shown in Fig. 6A. The symmetry matching technique detects this difference. Fig. 8C shows a negative largest excursion and an output that is not symmetric about this absolute maximum. A poor scoring factor M_C0EF(j) of zero would be as¬ signed.

3. Matching Against Dirac Pulse

Fig. 9 shows a technique matching against Dirac pulse that embodies features of the invention. This matching technique employs a whitening algorithm to first filter the template electrograms and the paced electrograms. The whitening filter transforms so-called colored noise, which can be 60- Hz (or 50-Hz) interference, or motion or muscular artifacts of the patient, to white noise.

The technique matched filters the whitened paced electrogram for each electrode using the left-right flipped, whitened template for that electrode as coefficients of the matched filter. Ideally, exactly matched, whitened electrograms will produce an output that equals a Dirac pulse. Therefore, each filter output is compared to a Dirac pulse. An algorithm scores the similarity for each electrode.

The pacing electrode whose whitened, matched filtered output most closely resembles a Dirac pulse is designated to be close to a potential ablation site. . Cross Correlation Technique

Fig. 10 shows a cross correlation technique that embodies features of the invention. This technique uses an appropriate algorithm to calculate for each electrode the cross correlation function between the template electrogram and the paced electrogram. For identical electrograms, the largest excursion of the cross correlation function will equal 1.0.

Various conventional methods for determining the cross correlation function can be used. For exam¬ ple, for M pairs of data {x(m) , y(m)}, where x(m) is the template electrogram and y(m) is the paced electrogram, the correlation function can be calcu¬ lated as follows:

where m = 1 to M; -M < k < M, and x and y are the means of the sequences {x} and {y}. ^M _COEF(Ϊ) ^^{s ec}3^ual ^{to the} largest excursion of the sequence {rxy(k)} computed for the individual electrode E(i) (i.e., the largest excursion can be either negative or positive, depending upon the degree of intercorrelation) . The pacing electrode Ep(j) having an overall matching factor A _{ACEI J )} closest to 1.0 is designated to be close to a potential ablation site. Additional information may be contained in the shift parameter k for each electrode. For example. Fig. 11A shows the cross correla¬ tion function for the electrograms of Fig. 6A and Fig. 6B. These electrograms are quite similar, and the cross correlation technique detects this. The largest excursion of the cross correlation function in Fig. 11A is near 1.0 (i.e., it is 0.9694).

Refer now to Fig. 11B, which shows the cross correlation function for the unlike electrograms shown in Figs. 6A and 8A. The cross correlation technique detects this lack of similarity. The largest excursion in Fig. 11B is negative (i.e., it is -0.7191) .

5. Norm of the Difference Technique Fig. 12 shows a norm of the difference technique that embodies features of the invention.

This technique normalizes, for each electrode, the template electrogram with respect to the abso¬ lute value of its largest excursion from baseline. This technique also normalizes, for each electrode, the paced electrogram with respect to the largest excursion from baseline. The technique then calcu- lates, for each electrode, the norm of the differ¬ ence between the template electrogram and the paced electrogram. The norm will decrease in proportion to the similarity of the electrograms.

For example, Fig. 13A is the difference between the similar electrograms shown in Figs. 6A and 6B, after each was normalized with respect to its largest excursion. This technique detects the similarity with a relatively small norm of the difference (i.e., it is 0.9620). Refer now to Fig. 13B, which is the difference between the dissimilar electrograms shown in Figs. 6A and 8A, after each was normalized with respect to its largest excursion. This technique detects the lack of similarity with a relatively high norm of the difference (i.e., it is 2.4972).

The technique preferably uses a weighted averag¬ ing algorithm to average, for each pacing electrode, the norm of the differences for all recording elec¬ trodes. The pacing electrode having the smallest average norm of the differences is designated the appropriate place to ablate.

The electrograms may or may not be filtered before analysis. A 1 to 300 Hz bandpass filter may be used for filtering. If a filter is used to reduce the noise for an electrogram that is used as a template, the same filter must also be used for the paced electrograms, since filtering may alter the electrogram morphology.

The electrograms might need to be aligned prior to processing. Any columnar alignment technique can be used. For example, the electrograms could be aligned about the point of largest positive slope.

The implementation of the system 10 described herein is based largely upon digital signal process- ing techniques. However, it should be appreciated that a person of ordinary skill in this technology area can easily adapt the digital techniques for analog signal processing.

The output signal y(t) of an analog matched filter is given by the analog convolution: y(t) - x(t) * Tflip(t); where Tflip(t) = EG(T-t) , which constitutes a left-right flipped replica of the electrogram template EG(t) that has the period T. Physically, an analog matched filter can be implemented with analog integrators and adders. Also, optical realizations of such filters can be implemented, for example, by using optical slots to represent the template. After optical conversion, the input signal is passed through the optical slot. The average light intensity behind the optical slot plane is maximal when the shape of the optically converted input signal matches the shape of the slot. An optical sensor can measure the average light intensity and output a signal that represents the matched coefficient M_C0EF(j;).

C. Location Output Isolation and Verification In one implementation, the host processor 52 sets a match target N_Mateh, which numerically estab- lishes a matching factor M_pACE(j) at which a high probability exists that the pacing electrode is close to a potential ablation site. In a preferred implementation, N^^ = 0.8. When M_p^^ > N_MTCH, the host processor 52 deems the location of the pacing electrode Ep(j) to be close to a potential site for ablation. When this occurs (as Fig. 4 shows) , the host processor 52 transmits a SITE signal to an associated output display device 54 (see Fig. 1A) . Through visual prompts, the display device 54 notifies the physician of the location of the pacing electrode Ep(j) and suggests that the location as a potential ablation site. When more two or more M_pACE * > N_MATCH, the host processor 52 sorts to locate the ^M _PACE(J₎ ^haviⁿ9 ^the highest value. In this instance, the host processor 52 deems the pacing electrode Ep(j) with the highest M_pACE(j) to be the one having the highest likelihood for being close to a potential ablation site and transmits the SITE signal accordingly. In the illustrated and preferred embodiment, the process controller 32 provides iterative pacing and matching using different pacing and matching tech¬ niques. Using different pacing-and-compare tech¬ niques allows the comparison of the location output from one technique with the location output from one or more different techniques. Using iterative pacing and matching, the process controller 32 and the host processor 52 confirm and cross-check the location output to verify its accuracy before ablation. The host processor 52 can also rely upon alternative diagnostic techniques to analyze the biopotential morphology.

In the illustrated and preferred embodiment (see

Fig. IB) , the system 10 also includes a roving pacing probe 68 usable in tandem with the basket structure 20 to generate and verify the location output.

1. Iterative Pacing In the illustrated and preferred embodiment (see Fig. IB) , the process controller 32 includes a module 60 that allows the physician to select among different types of pacing techniques. The different pacing techniques allow the physician to conduct both global and localized site identification, with and without inducing the abnormal cardiac event.

At least one technique is appropriate for pacing a large tissue region to identify a subregion close to a potential ablation site, without the need to induce an abnormal cardiac event. In a preferred implementation, the process controller 32 first conditions the pacing module 48 in the mode to conduct pace mapping. Pace mapping uses all elec¬ trodes 24 on the structure 20 in sequence as the pacing electrode, and does not induce the cardiac event. Based upon pace mapping, the process control¬ ler 32 obtains a location output that points to a general subregion that is close to a potential ablation site.

Once the general region is identified, another pacing technique can be employed to more narrowly define the location of the potential ablation site within the general region. In a preferred implemen¬ tation, the process controller 32 conditions the pacing module 48 in the matching mode to carry out entrainment or reset pacing, using the electrodes in the general subregion as the pacing electrodes. Entrainment or reset pacing in this subregion overdrives the arrhythmia, and provides enhanced differentiation of slow conduction zones. The process controller 32 thereby obtains a location output that is more localized with respect to the potential ablation site.

2. Iterative Matching

In the illustrated and preferred embodiment (see Fig. IB) , the process controller 32 also includes a module 62 that allows the physician to select more than one matching technique during iterative pacing.

For example, the process controller 32 may, during pace mapping or entrainment/reset pacing, compare the templates to the paced electrograms by first pattern matching, then by symmetry matching, and then by norm of the difference. In this way, the process controller 32 determines the uniformity of the location output among the different matching techniques. The correspondence of the location outputs confirms their reliability.

3. Cross-check Using Alternative Diagnos¬ tic Techniques

In the preferred embodiment, the process con- troller 32 is also electrically coupled by a bus 64 to a diagnostic module 66. Under the control of the process controller 32, as selected by the physician, the diagnostic module 66 conducts one or more alternative analyses of heart activity to cross check or verify the output location that the process controller 32 generates based upon electrogram matching.

In the illustrated embodiment (see Fig. IB) , the module 66 determines the fractionation of the paced electrograms. The degree of fractionation can be used as a cross-check that the physician can employ to cross-check and verify the output location or locations that the process controller 32 yields when operated in the matching mode. 4. The Roving Pacing Probe

Using the above iterative pacing and matching techniques, the location output may comprise a single electrode 24 or several electrodes 24 in a localized region of the structure 20. In the illus- trated and preferred embodiment (see Fig. IB) , the system 10 further includes a roving pacing probe 68 that can be deployed in the heart region 12 while the multiple electrode structure 20 occupies the region 12. The roving probe 68 is electrically coupled to the pacing module 48 to emit pacing signals.

In use, once the process controller 32 generates the output location or locations using the elec¬ trodes 24 to pace the heart, the physician positions the roving electrode probe 68 within the localized region near the output location electrode or elec¬ trodes 24. The process controller 32 preferably includes a homing module 70 to aid the physician in guiding the roving electrode probe 68 in the local- ized region within the structure 20. Systems and methods for operating the homing module 70 are dis¬ closed in copending patent application Serial No. 08/320,301, filed October 11, 1994, and entitled "Systems and Methods for Guiding Movable Electrode Elements Within Multiple Electrode Structures", which is incorporated herein by reference.

The process controller 32 conditions the pacing module 48 to emit pacing signals through the roving pacing probe 68 to pace the heart in the localized region, while the electrodes 24 record the resulting electrograms. By pacing this localized region with the roving pacing probe 68, while comparing the paced electrograms with the templates, the process controller 32 provides the capability of pacing and comparing at any location within the structure 20. In this way, the process controller 32 generates as output a location indicator that locates a site as close to a potential ablation site as possible. Of course, iterative pacing and matching techniques, as above described, can be practiced using the roving pacing probe 68.

Due to the often convoluted and complex contours of the inside surface of the heart, the basket structure 20 cannot contact the entire wall of a given heart chamber. The preferred implementation of the system 10 (as Fig. IB shows) therefore deploys the roving pacing probe 68 outside the structure to pace the heart in those wall regions not in contact with the electrodes 24. The roving pacing probe 68 can also be deployed while the basket structure 20 occupies the region 12 to pace the heart in a different region or chamber. In either situation, the electrodes 24 on the structure 20 record the resulting paced electrograms for comparison by the process controller 32 to the templates. The process controller 32 is thus able to generate an output identifying a location close to a potential ablation site, even when the site lies outside the structure 20 or outside the chamber that the structure 20 occupies.

Acting upon the location output generated in accordance with the invention, the physician deploys the ablation electrode 36 to the location of the pacing electrode Ep(j) to conduct the ablation (as Fig. 1A shows) . The homing module 70 (as already - 37 -

described and as shown in Fig. IB) can also be used to aid the physician in deploying the ablation elec¬ trode 36 to the designated site, as disclosed in copending patent application Serial No. 08/320,301, filed October 11, 1994, and entitled "Systems and Methods for Guiding Movable Electrode Elements Within Multiple Electrode Structures", which is incorporated herein by reference.

It should be appreciated that the system 10 is not limited to the diagnosis and treatment of arrhythmia events. The system 10 can be used in the sampling mode, for example, to create templates while the heart is in sinus rhythm. In the matching mode, the heart can be paced at sinus rhythm rates and the paced electrograms compared to the templates to detect abnormal activation patterns associated with other forms of heart disease or to identify the presence of accessory pathways. D. Time-Sequential Analysis In the illustrated and preferred embodiment, the template electrograms at all electrodes 24 are recorded during the same time interval. Likewise, the pacing electrograms for each pacing electrode Ep(j) are recorded at all electrodes 24 during the same time interval. This technique requires the process controller 32 to have parallel processing channels equal in number to the number of electrodes 24 conditioned to record the electrograms.

For example, it is typically desired to record electrogram information from thirty-two (32) elec¬ trode pairs when analyzing monomorphic VT. When the electrograms are recorded over the same time inter¬ val, the process controller 32 must be capable of handling thirty-two (32) parallel channels of information. In an alternative embodiment, the process controller 32 can be operated in a time-sequential recording mode. In this mode, the process control¬ ler 32 records electrograms, either to create a template or to create paced electrograms for match¬ ing, at different time intervals. In this mode, the process controller 32 consolidates the time-sequen¬ tial electrograms for composite analysis, as if the electrograms were recorded during the same time intervals.

The time-sequential mode can be used when the waveshapes of the electrograms to be analyzed are generally the same during each heart beat. For example, monomorphic VT is characterized by such time-invariant electrogram waveshapes. The time- sequential mode allows the physician to condition the process controller 32 to record time invariant electrograms in numbers greater than the number of processing channels that the process controller 32 has.

For example, if the process controller 32 can only accommodate twenty (20) channels of data at a given time, the time-sequential mode nevertheless allows information from thirty-two (32) electrograms to be recorded and processed.

Fig. 21 shows the time-sequential mode of operation in diagrammatic flow chart form. During a first time interval TI(1), the time-sequential mode simultaneously records electrograms at first elec- trode sites ES(1) . In Fig. 21, the first electrode sites ES(1) number twenty (20) and are designated E(l) to E(20) . The electrograms for the first site electrodes E(l) to E(20) are retained for the first time interval TI(1). Fig. 22A shows representative electrograms recorded at E(l) during TI(1). Fig. 22B shows a representative electrogram recorded at E(2) during TI(1) .

During a second time interval TI(2), the time- sequential mode simultaneously records electrograms at second electrode sites, at least one of which is an electrode site used during the first time inter¬ val TI(1). Fig. 21 identifies E(l) as the common electrode site. E(21) to E(32) comprise the remain¬ ing electrodes in ES(2). Fig. 22C shows representa- tive electrograms recorded at the common electrode site E(l) during TI(2). Fig. 22D shows representa¬ tive electrograms recorded at the additional elec¬ trode site E(21) during TI(2).

In a preferred implementation, the process controller 32 conditions the first electrode sites ES(1) for recording and records electrograms from these sites for TI(1) = 4 seconds. The process controller 32 then automatically conditions the second electrode sites ES(2) for recording and records electrograms from these sites for TI(2) = 4 seconds. Thus, over a time-sequenced interval of eight (8) seconds, the process controller 32 has recorded electrograms at thirty-two (32) electrode sites, which is the number desired for analysis. The time-sequential mode determines the time difference TD between the electrograms for E(l) at TI(1) and TI(2). Fig. 22C shows TD.

The time-sequential mode time-aligns E(l) at TI(2) with E(l) at TI(1) by left-shifting E(l) at TI(2) by TD. Fig. 23C shows the representative electrograms recorded at E(l) during TI(2) after time-alignment with the electrograms recorded at E(l) during TI(1), which are shown in Fig. 23A.

The time-sequential mode also left-shifts the electrograms E(21) through E(32) by the same amount TD. Fig. 23D shows the representative electrograms recorded at E(21) during TI(2) after time-alignment. In this way, the time-sequential mode creates the electrogram composite EC, which consists of the time-registered electrograms E(l) to E(32) taken at TI(1) and TI(2). The time-alignment process of creating the electrogram composite EC can be done manually by the physician, by interacting with the display device 54. Preferably, the host processor 52 automatically analyzes the signals, computes TD, and accomplishes the time-alignment to create the composite electrogram EC. Alternatively, TD need not be computed. The physician or the operator can make use of any time-assignment method to align the signals based on the information contained in the common channels E(l) . An example of useful informa¬ tion is the location of the maximal slopes of E(l) . Of course, this algorithm can be automatically implemented and executed. Though taken at different time intervals TI(1) and TI(2), the time-aligned electrogram composite EC can be analyzed in the same manner as electrograms taken simultaneously during the same time interval. In the context of the system 10 described herein, the electrogram composite EC can be used to create the electrogram template, or to create paced electrograms for comparison with an electrogram template, or as electrograms for any other diagnos¬ tic purpose. For example, using the above described time- sequential methodology, virtually all types of signals derived from biological events can be processed, such as electrocardiograms, tissue bipotential signals, pressure waves, electrogastrograms, electromyograms, electroen- cephalograms, impedance measurements, and tempera¬ ture measurements.

11. Endocardially Paced Electrocardiograms A. Electrocardiogram Matching In the preceding embodiments, the endocardially positioned basket structure 20 both paces and senses the resulting electrograms. In an alternative implementation, the process controller 32 can condition the pacing module 48 in the sampling mode to pace the heart to induce a desired cardiac event, using individual or pairs of electrodes 24 on the basket structure 20 deployed in the heart region 12 (as already described) , while creating templates of the resulting electrocardiograms recorded by the processing module 50 from body surface electrodes electrically coupled to the process controller 32.

In this implementation, during the matching mode, the process controller 32 paces the heart with the individual or pairs of endocardial electrodes 24 positioned on the structure 20 in the heart region

12. The resulting paced electrocardiograms are recorded by the same body surface electrodes (locat¬ ed in the same position as during the sampling mode) and compared to the electrocardiogram templates in the manner above described.

In this implementation, the process controller 32 generates the location output based upon compar¬ ing the electrocardiogram sample templates with endocardially paced electrocardiograms. B. Electrocardiogram Time Delays

Endocardially paced electrocardiograms can also be used to identify regions of slow conduction.

In this implementation, while the process controller 32 conditions the pacing module 48 to pace the heart with the individual or pairs of electrodes 24 positioned on the structure 20 endocardially in the heart region 12, the resulting endocardially paced electrocardiograms are recorded by body surface electrodes coupled to the process controller 32. From the endocardially paced elec¬ trocardiograms, the process controller 32 measures the time difference between the pacing signal and the onset of the Q-wave to detect slow conduction regions (characterized by abnormally large time delays) .

Preferably, the process controller 32 generates maps displaying iso-time delay regions based upon these endocardially paced electrocardiograms, to further aid in the location of the slow conduction region.

C. Characterizing Tissue Morphology Time delays obtained from endocardially paced electrocardiograms can also characterize heart tissue morphology. In this implementation, the body surface elec¬ trodes record electrocardiograms while the pacing module 48 paces the heart with the individual or pairs of electrodes 24 positioned on the structure 20 in the heart region 12. The pacing module 48 first paces the heart at or near normal sinus rhythm rates. The process controller 32 registers the time delays recorded from the resulting electrocardio¬ grams. The pacing module 48 next paces the heart at an increased rate, e.g., at or near an arrhythmia rate. The process controller 32 registers the resulting time delays from the resulting electro¬ cardiograms.

The process controller 32 compares the paced sinus rate time delays with the paced arrhythmia rate time delay. The location of the pacing elec- trodes where the time delays shortened as the pacing rate increased are near regions of healthy tissue. The location of pacing electrodes where the time delays lengthened as the pacing rate increased are near regions of ischemic tissue. The process controller 32 preferably generates iso-display maps showing the distribution of the time delay differ¬ ences, thereby aiding the physician in differentiat¬ ing between regions of healthy and ischemic tissue. III. Pacing Artifact Removal

Pacing artifacts in the pacing electrograms may be eliminated by conventional techniques to better discern the initial point of depolarization. Howev¬ er, in the illustrated and preferred embodiment, the process controller 32 includes a filter 56 (see Fig. 4) that removes the pacing artifact for this purpose without otherwise altering the morphology of the electrogram. The operation of the filter 56 may vary. A. Nonlinear Filter

In the preferred embodiment, the filter 56 implements a nonlinear sorting algorithm of the type shown in Fig. 14A.

Fig. 14B shows a representative implementation and filter output for the algorithm in diagrammatic form.

The algorithm establishes a sample window. The sample window has a predetermined length (WL) , ex¬ pressed in terms of the number of discrete sample points the window contains. The predetermined length (WL) of the sample window takes into account the length (AL) of the pacing artifact, which is ex¬ pressed in terms of the number of sample points that encompass the pacing artifact. Preferably, the window length WL is an odd number. If WL is significantly smaller than twice AL, the sorting algorithm will not serve to eliminate the pacing artifact to the extent necessary to accurately discern the initial point of depolariza- tion. There is, however, a limitation placed upon how large WL is relative to the size of AL. When WL is significantly larger than twice AL, the morpholo¬ gy of the electrogram will be distorted by being spread out with respect to time. It is believed that the sample window length should preferably be at least twice the pacing artifact length. Most preferably, WL > 2AL + k, where k = 1 to WL/2. k is an additive amount that optimizes the elimination of the artifact without time distortion.

The algorithm advances the sample window along the electrogram, taking a succession of boxcar sam¬ ples, X(n), where n = 1 to WL. The algorithm sorts the sample values X(n) in the window from smallest value to largest value. The sorted permutation is the sequence {X(p[n])}, where {p[n]} is a permuta¬ tion of {n} resulting from the sorting process, and where n = 1 to WL. The algorithm selects one of the sort positions p[f] according to prescribed crite- ria, where f = 1 to WL. The selection criteria will be discussed in greater detail later.

The algorithm outputs the sample value X(p[f]) contained in the selected sort position, which constitutes the filter output for the boxcar sample. The algorithm outputs X(p[f]) and advances the window forward in time one sample point.

The algorithm repeats the sorting process, generating a filter output for each boxcar sample and advancing the window, until the entire electro- gram has been processed. The algorithm then plots the filter outputs with respect to time, which constitutes the filtered electrogram.

EXAMPLE (Nonlinear Filtering) For example, given WL = 5, the sequence of samples values X(n) is:

1 X(n) X ( l ) X ( 2 ) X ( 3 ) X ( 4 ) X ( 5 )

[value 4 2 10 8 6

The sequence of sample values X(l to 5) consti¬ tutes the boxcar sample.

The algorithm sorts the sequence X(l to 5) in increasing numerical value, or

X(2) < X(l) < X(5) < X(4) < X(3).

The algorithm establishes the sort positions p(n) based upon this permutation, or

X(p(n)) X(P(1)) X(p(2» X(P(3)) X(P(4)) X(p(5»

Value 2 4 6 8 10

The algorithm selects a sort position p(f) according to prescribed criteria. The criteria for selecting the sort position takes into account the length of the artifact AL, as will be discussed later. In the preferred embodiment, the criteria specifies the sort position relative to the other sort positions. In this implementation, (f) is ex¬ pressed as a position (z) of WL positions, i.e., p(z/WL), where WL is the size of the sort window. The position z is selected taking into account AL, and, more particularly, z should increase as AL decreases.

For example, for WL = 5, if z = 3, then p(3/5) means that X(p(3)) replaces x(3) in the output se¬ quence. The value X(p(3)) is 6, which becomes the filter output for this boxcar sample, based upon the selected sort position criteria.

In this particular case, p(3/5) criteria actual¬ ly implements a median filter. For a given window, the following sort position z constitutes the median

-1 ,—WL+ \ .1

where:

1 < z < WL, and the expression:

, WL+\ *.

represents the integer part of:

WL

2 For example [3.9] = 3, just as [3.1] = 3.

Further details on median filtering techniques are disclosed in "VLSI Array Processors" by S. Y. Kung (Prentice Hall, (1991).

Alternatively, if the selected sort position is p(4/5), the value X(p(4)) is 8, which becomes the nonlinear, non-median filter output for this boxcar sample, based upon the selected sort position criteria. This corresponds to x(4) of the output sequence. The algorithm advances the window one sample at a time, sorting the sample enclosed within the window, and generating a filter output based upon the sort criteria, and so on until the entire electrogram has been filtered. In the preferred embodiment, the algorithm keeps the timing of filter output in sequence with the timing of the electrogram by retaining the value of edge samples, so that the number of filter outputs equal the number of electrogram samples. The number of edge values retained, of course, depends upon the size of the sample window WL.

More particularly, the algorithm retains a prescribed number, y., of beginning sample values a number, y₂, of ending sample values, arranging the filter output between the prescribed number of beginning and ending sample values to keep the filter output arranged with respect to time in sequence with the derived biological signal. In the preferred implementation, y, =z- \

and y₂= WL-z

Fig. 14B shows the filtering of ten sample points (4 2 7 5 1 10 3 8 9 6) in accordance with the above described technique. The window length WL in Fig. 14B is 5, and the sort criteria is median filtering, i.e. p(3/5) . Fig. 14B shows the reten- tion of the edge samples, two samples (4 and 2) at the front edge (y., = z - l or 3 - l = 2) and two samples 9 and 6 at the rear edge (y₂ = WL - z or 5 - 3 = 2). Fig. 14B also shows the filter outputs (4 5 5 5 8 8) between the edge samples, with the sorted samples appearing to the right of the filter outputs.

Fig. 14C shows the filtering of the same ten sample points (4 2 7 5 1 10 3 8 9 6), with the same window length WL of 5, but with a non-median sort criteria p(4/5) . Fig. 14C shows the retention of the edge samples, three samples (4, 2, 7) in at the front edge (y₁ = z - 1 or 4 - 1 = 3) and one sample 6 at the rear edge (y₂ = WL - z or 5 - 4 = 1) . Fig. 14C shows the filter output for the p(4/5) criteri¬ on: (5 7 7 8 9 9) between the edge samples.

The selection of the sort position p(f) takes into account the morphology of the pacing artifact in terms of the length of the artifact AL, expressed in terms of the number of sample points that encom¬ pass it. The percentage value of f should increase as the artifact length AL decreases, or, given a constant WL, z should increase as AL decreases. EXAMPLE (Sort Position Selection Criteria) Fig. 15A shows a simulated pacing artifact where AL is 5 and the width of the highest peak is 3. Fig. 15B shows the filtered output for p(2/5) ; Fig. 15C shows the filtered output for the median, or p(3/5); and Fig. 15D shows the filtered output for p(4/5). The criteria p(2/5) fully eliminated the pacing artifact (Fig. 15B) , whereas the criteria p(3/5) and p(4/5) did not (Figs. 15C and 15D, respectively) . Thus, the optimal elimination of certain pacing artifacts requires nonlinear, non- median filtering, where the position z comprises a positive integer; 1 < z < WL; and:

, WL+ \ _Λ *[— ^~—1

Fig. 16A shows a simulated pacing artifact where

AL is 5 and the width of the highest peak is 2. Fig. 16B shows the filtered output for p(2/5) ; Fig.

16C shows the filtered output for p(3/5), i.e. the median; and Fig. 16D shows the filtered output for p(4/5). The criteria p(3/5) fully eliminated the pacing artifact (Fig. 16C) , whereas the criteria p(2/5) and p(4/5) did not (Figs. 16B and 16D, respectively) . Fig. 17A shows a simulated pacing artifact where AL is 5 and the width of the highest peak is 1. Fig. 17B shows the filtered output for p(2/5); Fig. 17C shows the filtered output for p(3/5) , i.e. the median; and Fig. 17D shows the filtered output for p(4/5). The criteria p(4/5) fully eliminated the pacing artifact (Fig. 17D) , whereas the criteria p(2/5) and p(3/5) did not (Figs. 17B and 17C, respectively) .

Fig. 18A shows a paced electrogram consisting of 500 samples taken in increments of 0.5 seconds. The designation PA in Fig. 18A marks the location of the pacing artifact, which is 5 sample periods in length (i.e., AL = 5) .

Fig. 18C shows a plot of the filter output generated by filtering the electrogram of Fig. 18A using a sample window size WL=7 (that is, less than twice the artifact sample size) and specifying the median p(4/7) as the sort position. Fig. 18C shows a reduction in the size but not an elimination of the pacing artifact PA by median filtering.

Fig. 18B shows a plot of the filter output generated by filtering the electrogram of Fig. 18A using a sample window size WL=11 (that is, WL = 2AL + 1) , while still specifying the median p(6/ll) as the sort position. Fig. 18B shows an elimination of the pacing artifact by median filtering.

Except for the median filter p(3/5) , the imple¬ mentation of this type of nonlinear filter will distort both the positive and negative phases of the useful signal (see Figs. 15B/D, 16B/D, and 17B/D.) B. Adaptive Filtering

Alternatively, the filter 56 can implement the adaptive filtering algorithm shown in Fig. 19 to remove the pacing artifact.

The filter 56 generates an internal variable TPACE(t) expressing a template of the pacing arti¬ fact itself. The function TPACE(t) preferably begins with a preestablished template typical for a pacing artifact. Alternatively, the algorithm can create an initial template by manually selecting a window about the artifact and creating a template by, for example, conventional signal averaging techniques. This template could be adaptively updated using appropriate signal averaging techniques.

Figs. 24 to 28 exemplify a preferred way of generating a template of the pacing artifact. Fig. 24 shows a portion of a recording of a paced elec¬ trogram extending over about two beats and a half. Fig. 24 shows three pacing pulses (PA-1; PA-2; PA- 3) , which are the artifacts that are to be ultimate¬ ly removed. The physician preferably selects windows about each pacing pulse PA-1 to 3 to create an averaged template. Alternatively, the physician can select one of the pacing pulse, for example PA- 1, to generate the template.

Figs. 25A to C represent the signals PS-l; PS-2; and PS-3 contained by the three windows manually selected about the pacing pulses, respectively PA-1; PA-2; and PA-3. The physician manually aligns the three signals PS-l; PS-2; and PS-3 and truncates them at the same length, as Figs. 26A to C show. In Figs. 26A to C, the three signals PS-l; PS-2; and PS-3 have been aligned about their largest positive peak, although other alignment techniques could be used.

Fig. 27 shows the template TPACE(t) of the pacing artifact generated by averaging the three signals PS-l; PS-2; and PS-3 after alignment and truncation (i.e., the signals shown in Figs. 26A to C are averaged) . As Figs. 28A and B show, the template TPACE(t) (Fig. 28B) is aligned with the first pacing pulse in the electrogram (Fig. 28A) prior to executing the adaptive algorithm for artifact removal.

It is not necessary to generate a new pacing artifact template TPACE(t) for the electrogram sensed by each electrode. The same initial template TPACE(t) from only one of the electrodes can be used for every electrogram. Alternatively, pacing signals from different electrodes can be aligned for averag¬ ing to create the template TPACE(t) , which is then used for all electrograms.

The template TPACE(t) can also be generated by approximating the pacing pulse using suitable mathematical techniques, for example, spline inter¬ polation. A universal template TPACE(t) can also be generated from recordings taken from different patients at different times with different equip- ment, although such different records may require proper adjustment before generating the universal template TPACE(t).

The filter 56 expresses the input signal with respect to time IN(t) in term of the function expressed as:

IN(t) = EG(t) + PACE(t) where:

EG(t) is the actual electrogram, and PACE(t) is the pacing artifact. The template TPACE(t) of the filter 56 reduces IN(t), so the output signal EG(t) is expressed as IN(t) - TPACE(t). The filter 56 changes TPACE(t) over time based upon the energy of the output EG(t) so as to minimize the energy of (PACE(t) - TPACE(t)) over time, and therefore, the energy of EG(t) .

Expressed differently, the filter 56 seeks to minimize the function EG(t) + PACE(t) - TPACE(t) over time. Ideally, the energy of EG(t) is mini¬ mized over time when TPACE(t) equals PACE(t) , therefore being equal to the energy of EG(t) .

The filter algorithm changes TPACE(t) over time applying known iterative techniques. For example, when applying the Least-Mean-Squares (LMS) tech¬ nique, the template for TPACE(t) is used as a refer- ence input for the LMS algorithm. The weight vector is initialized at [K 0 0 ... 0]. The variable K is chosen equal to the ratio between the peak of PACE(t) and the peak of the template TPACE(t). When PACE(t) = TPACE(t) , k is equal to one. Further details of LMS are found in "Adaptive Filter Theory" by S. Haykin (Prentice Hall, 1991)

Other conventional iterative techniques like Recursive-Least-Squares or Steepest-Descent can also be used to achieve the same result. EXAMPLE (Adaptive Filtering)

Fig. 20A shows a representative paced electro¬ gram. The designation PA in Fig. 20A marks the location of the pacing artifact. Fig. 2OB shows the paced electrogram after filtering using the LMS technique above described. Fig. 20B shows the effectiveness of the adaptive filter to remove the pacing artifact, without otherwise altering the morphology of the paced electrogram.

Either of the above described techniques for removing the pacing artifact have application outside the conditioning of the electrogram for morphology matching described herein. Either tech¬ nique has application whenever it is desired to remove an artifact signal from a useful signal or to otherwise eliminate virtually any signal of a known shape.

As a general proposition, nonlinear filtering or adaptive filtering, as above described, can be used whenever it is desired to remove cardiac related or other periodic artifacts, for example, in respirato¬ ry signals, or EEG's, or from neurological signals. Nonlinear filtering or adaptive filtering, as above described, can also be used to eliminate periodic artifacts that are not cardiac related, for example, 50 to 60 Hz noise from sensed signals due to poor power source isolation.

In the presence of a pacing artifact, nonlinear filtering or adaptive filtering, as above described, can also be used to remove the pacing artifact before measuring the level of fractionation in an electrogram. Since a pacing artifact looks much like an electrogram, it is desirable to remove it before analyzing for actual fractionation.

As another example, nonlinear filtering or adap- tive filtering, as above described, can be used to remove a pacing artifact when it is desired to conduct a frequency domain analysis of the cardiac signal, to determine the regularity of the heart beat. Any portion of the electrogram can be isolated for elimination using the filtering techniques de¬ scribed above, not merely the pacing artifact. The nonlinear and adaptive filtering techniques can be used in applications where low pass filtering cannot be used. For example, while body surface mapping can use low pass filtering of electrograms, endocar¬ dial mapping cannot, due to the use of higher fre¬ quencies than in electrocardiograms. For example, a common electrocardiogram frequency spectrum is .05 to 100 Hz, whereas a common bipolar electrogram spectrum is 1 to 300 Hz.

Nonlinear filtering or adaptive filtered, as above described, can be used for processing or analyzing virtually any signal derived from a biological event. In addition to processing or analyzing signals derived from cardiac-related events, nonlinear filter or adaptive filtering can be used to process or analyze electroencephalograms, respiratory signals, electrogastrograms, and electromyograms.

Various features of the invention are set forth in the following claims.

Claims

We claim:

1. An analog or digital processing element for analyzing biopotential morphologies in body tissue comprising first means for inputting a template of a biopotential event in body tissue, second means for inputting a sample of a biopotential event externally triggered in body tissue, and third means for electronically comparing the input sample to the input template and generat¬ ing an output based upon the comparison.

2. An analog or digital processing element for analyzing biopotentials in myocardial tissue comprising first means for inputting a template of a cardiac event of known diagnosis, second means for inputting a sample of a paced cardiac event, and third means for electronically comparing the input sample to the input template and generating an output based upon the comparison.

3. An analog or digital processing element for analyzing biopotentials in myocardial tissue comprising first means for inputting a template of an electrogram having a duration not greater than one heart beat recorded during a cardiac event of known diagnosis, second means for inputting a sample of a paced electrogram having a duration not shorter than the template, and third means for electronically comparing the input sample to the input template and generating an output based upon the comparison.

4. An analog or digital processing element for analyzing biopotentials in myocardial tissue comprising first means for inputting a template of an electrogram having a duration not greater than one heart beat recorded during a cardiac event of known diagnosis, second means for inputting a sample of a paced electrogram having a duration not shorter than the template recorded during entrainment or reset pacing, and third means for electronically comparing the input sample to the input template and generating an output based upon the comparison.

5. An element according to claim 1 or 2 or 3 or 4 wherein the output comprises a matching coefficient indicating how alike the input sample is to the input template.

6. An element according to claim 1 or 2 or 3 or 4 wherein the third means compares the input sample to the input template by matched filtering.

7. An element according to claim 1 or 2 or 3 or 4 wherein the third means compares the input sample to the input template by cross correlation.

8. An element according to claim 1 or 2 or 3 or 4 wherein the third means compares the input sample to the input template by deriving a norm of the difference.

9. An element according to claim 1 or 2 or 3 or 4 , wherein the third means compares the input sample to the input template by using the input template to create a matched filtered sample and by analyzing the symmetry of the matched filtered sample.

10. A system for analyzing biopotential morphologies in body tissue comprising electrode means for sensing biopotentials in body tissue, and an analog or digital processing element coupled to the electrode means comprising first means for conditioning the elec¬ trode means to sense a sample of biopotentials occurring in the tissue region during a biopotential generating event of known diagnosis and for creating a template based upon the sensed biopotential sample, second means for conditioning the electrode means to emit a signal to trigger biopotentials in body tissue and to sense using the electrode means a sample of the triggered biopotentials occurring in the tissue region as a result of the signal, and third means for electronically com- paring the triggered biopotential sample to the template and generating an output based upon the comparison.

11. A system according to claim 10 and further including an element supporting the electrode means for operative association with a region of body tissue.

12. A system according to claim 10 wherein the output comprises a matching coefficient indicating how alike the triggered biopotential sample is to the template.

13. A system according to claim 10 wherein the third means compares the triggered biopotential sample to the template by matched filtering.

14. A system according to claim 10 wherein the third means compares the triggered biopotential sample to the template by cross correlation.

15. A system according to claim 10 wherein the third means compares the triggered biopotential sample to the template by deriving a norm of the difference.

16. A system according to claim 10 wherein the third means compares the triggered biopotential sample to the template by using the template to create a matched filtered triggered biopotential sample and by analyzing the symmetry of the matched filtered triggered biopotential sample.

17. A system for analyzing biopotential morphologies in myocardial tissue comprising electrode means for sensing biopotentials in myocardial tissue, and an analog or digital processing element coupled to the electrode means comprising first means for conditioning the elec¬ trode means to sense a sample of the biopotentials occurring in the tissue region during a cardiac event of known diagnosis and for creating a template based upon the sensed biopotential sample, second means for conditioning the electrode means to emit a pacing signal and to sense using the electrode means a sample of the paced biopotentials occurring in the tissue region as a result of the pacing signal, and third means for electronically com- paring the paced biopotential sample to the template and generating an output based upon the comparison.

18. A system according to claim 17 and further including an element supporting the electrode means for operative association with a region of heart tissue.

19. A system for analyzing biopotential morphologies in myocardial tissue comprising electrode means for sensing biopotentials in myocardial tissue, an element supporting the electrode means for operative association with a region of heart tissue, and an analog or digital processing element coupled to the electrode means comprising first means for conditioning the elec¬ trode means to sense a sample of biopotentials occurring in the tissue region during a cardiac event of known diagnosis and for creating a template based upon the sensed biopotential sample, second means for conditioning the electrode means to emit a pacing signal and to sense using the electrode means a sample of the paced biopotentials occurring in the tissue region as a result of the pacing signal, and third means for electronically com¬ paring the paced biopotential sample to the template and generating an output based upon the comparison.

20. A system according to claim 17 or 19 wherein the output comprises a matching coefficient indicating how alike the paced biopotential sample is to the template.

21. A system according to claim 17 or 19 wherein the third means compares the paced biopotential sample to the template by matched filtering.

22. A system according to claim 17 or 19 wherein the third means compares the paced biopotential sample to the template by cross correlation.

23. A system according to claim 17 or 19 wherein the third means compares the paced biopotential sample to the template by deriving a norm of the difference.

24. A system according to claim 17 or 19 wherein the third means compares the paced biopotential sample to the template by using the template to create a matched filtered paced biopotential sample and by analyzing the symmetry of the matched filtered paced biopotential sample.

25. A system according to claim 17 or 19 wherein the sensed biopotential sample of the template comprises an electrogram of a first predetermined duration, and wherein the sensed paced biopotential sample comprises an electrogram of a second predetermined duration not shorter than the first predetermined duration.

26. A system according to claim 25 wherein the first and second predetermined durations are equal.

27. A system according to claim 26 wherein the first and second predetermined durations comprise one or a portion of one heart beat.

28. A system for analyzing biopotential morphologies in myocardial tissue comprising multiple electrode means for sensing biopotentials in myocardial tissue, an element supporting the multiple electrode means for operative association with a region of heart tissue, and an analog or digital processing element coupled to the multiple electrode means comprising first means for conditioning the multiple electrode means to sense a sample of the biopotentials occurring in the tissue region during a cardiac event of known diagnosis and for creating a template based upon the sensed biopotential sample, second means for pacing the heart at a predetermined location while sensing with the multiple electrode means a sample of the paced biopotentials occurring in the tissue region as a result of pacing, and third means for electronically com¬ paring the paced biopotential sample to the template and generating an output based upon the comparison.

29. A system according to claim 28 wherein the element supports the multiple electrode means in a three dimensional, radially expanded array.

30. A system according to claim 28 wherein the output comprises a matching coefficient indicating how alike the paced biopotential sample is to the template.

31. A system according to claim 28 wherein the second means changes the predetermined location for pacing the heart while sensing with the multiple electrode means a sample of the paced biopotentials occurring as a result of pacing at each location, and wherein the third means individually compares ,the paced biopotential samples for each pacing location to the template and generates an output associated with each pacing location.

32. A system according to claim 31 wherein each output comprises a matching coefficient indicating, for each pacing location, how alike the paced biopotential sample is to the template.

33. A system according to claim 28 or 31 wherein the third means compares the paced biopotential sample to the template by matched filtering.

34. A system according to claim 28 or 31 wherein the third means compares the paced biopotential sample to the template by cross correlation.

35. A system according to claim 28 or 31 wherein the third means compares the paced biopotential sample to the template by deriving a norm of the difference.

36. A system according to claim 28 or 31 wherein the third means compares the triggered biopotential sample to the template by using the template to create a matched filtered triggered biopotential sample and by analyzing the symmetry of the matched filtered triggered biopotential sample.

37. A system for analyzing biopotential morphologies in myocardial tissue comprising an array of multiple electrodes, an element supporting the multiple electrode array within a heart chamber in operative association with a region of endocardial tissue, and an analog or digital processing element coupled to the multiple electrode means comprising first means for conditioning the multiple electrodes on the array to sense at each electrode a sample of the biopotentials occurring in the tissue region during a cardiac event of known diagnosis and for creating a template for each electrode based upon the sensed biopotential sample, second means for conditioning at least one of the multiple electrodes on the array to emit a pacing signal and to sense at each electrode on the array a sample of the paced biopotentials occurring in the tissue region as a result of the pacing signal, and third means for electronically com¬ paring the paced biopotential sample sensed at each electrode to the template for that electrode and generating an output for each electrode based upon the comparison.

38. A system according to claim 37 wherein the output for each electrode comprises a matching coefficient indicating, for each electrode, how alike the paced biopotential sample is to the template.

39. A system according to claim 37 wherein the second means sequentially conditions different ones of the multiple electrodes on the array to emit a pacing signal and senses at each electrode on the array a sample of the paced biopotentials occurring as a result of the pacing at each different one of the electrodes, and wherein the third means individually compares, for each different one of the pacing signal-emitting electrodes, the paced biopotential samples sensed by each multiple electrode to the templates for each multiple electrode and generates an output associated with each multiple electrode.

40. A system according to claim 39 wherein the output for each multiple electrode comprises a matching coefficient indicating how alike the paced biopotential sample is to the template.

41. A system according to claim 40 wherein the third means further derives, for each pacing signal-emitting electrode, an overall matching factor from the matching coefficients for each multiple electrode, the overall matching factor indicating how alike all the paced biopotential samples are to the templates.

42. A system according to claim 37 or 39 wherein the third means compares the paced biopotential sample to the template by matched filtering.

43. A system according to claim 37 or 39 wherein the third means compares the paced biopotential sample to the template by cross correlation.

44. A system according to claim 37 or 39 wherein the third means compares the paced biopotential sample to the template by deriving a norm of the difference.

45. A system according to claim 37 or 39 wherein the third means compares the paced biopotential sample to the template by using the template to create a matched filtered paced biopotential sample and by analyzing the symmetry of the matched filtered paced biopotential sample.

46. A system according to claim 37 or 39 wherein the sensed biopotential sample of the template comprises an electrogram of a first predetermined duration, and wherein the sensed paced biopotential sample comprises an electrogram of a second predetermined duration not shorter than the first predetermined duration.

47. A system according to claim 46 wherein the first and second predetermined durations are equal.

48. A system according to claim 46 wherein the first and second predetermined durations comprise one or a portion of one heart beat.

49. An analog or digital processing element for analyzing biopotentials in myocardial tissue to locate potential ablation sites comprising first means for inputting a template of a cardiac event of known diagnosis, second means for inputting a sample of a paced cardiac event triggered from a known pacing site, third means for electronically comparing the input sample to the input template and generating an output based upon the comparison, the output comprising a matching factor indicating how alike the input sample is to the input template, and fourth means for comparing the matching factor to a predetermined value to determine whether the pacing site is a potential ablation site.

50. An analog or digital processing element for analyzing biopotentials in myocardial tissue to locate potential ablation sites comprising first means for inputting a template of an electrogram recorded during a cardiac event of known diagnosis, second means for inputting a sample of a paced electrogram triggered from a known pacing site, third means for electronically comparing the input sample to the input template and generating an output based upon the comparison, the output comprising a matching factor indicating how alike the input sample is to the input template, and fourth means for comparing the matching factor to a predetermined value to determine whether the pacing site is a potential ablation site.

51. An analog or digital processing element for analyzing biopotentials in myocardial tissue to locate potential ablation sites comprising first means for inputting a template of an electrocardiogram recorded during a cardiac event of known diagnosis, second means for inputting a sample of a paced electrocardiogram triggered from a known pacing site, third means for electronically comparing the input sample to the input template and generating an output based upon the comparison, the output comprising a matching factor indicating how alike the input sample is to the input template, and fourth means for comparing the matching factor to a predetermined value to determine whether the pacing site is a potential ablation site.

52. An element according to claim 49 or 50 or 51 wherein the third means compares the input sample to the input template by matched filtering.

53. An element according to claim 49 or 50 or 51 wherein the third means compares the input sample to the input template by cross correlation.

54. An element according to claim 49 or 50 or 51 wherein the third means compares the input sample to the input template by deriving a norm of the difference.

55. An element according to claim 49 or 50 or 51 wherein the third means compares the input sample to the input template by using the input template to create a matched filtered sample and by analyzing the symmetry of the matched filtered sample.

56. An element according to claim 49 or 50 or 51 wherein the fourth means provides an output when the matching factor exceeds the predetermined value.

57. An analog or digital processing element for analyzing biopotential morphologies in myocardial tissue to locate potential ablation sites comprising first means for inputting a template of a cardiac event of known diagnosis in a region of myocardial tissue, second means for inputting a first sample of paced biopotentials occurring in the tissue region as a result of pacing using a first technique at a known pacing site, third means for inputting a second sample of paced biopotentials occurring in the tissue region as a result of pacing at the known pacing site using a second technique different than the first technique, fourth means for electronically comparing each of the first and second samples individually to the template and generating first and second outputs based upon each comparison, each output comprising a matching factor indicating how alike the input sample is to the input template, and fifth means for comparing at least one of the matching factors to a predetermined value to determine whether the pacing site is a potential ablation site.

58. An element according to claim 57 wherein the third means compares the input sample to the input template by matched filtering.

59. An element according to claim 57 wherein the third means compares the input sample to the input template by cross correlation.

60. An element according to claim 57 wherein the third means compares the input sample to the input template by deriving a norm of the difference.

61. An element according to claim 57 wherein the third means compares the input sample to the input template by using the input template to create a matched filtered sample and by analyzing the symmetry of the matched filtered sample.

62. An element according to claim 57 and further including sixth means for comparing the first and second matching factors.

63. An element according to claim 57 wherein the fifth means provides an output when at least one of the matching factors exceeds the predetermined value.

64. An element according to claim 63 wherein the fifth means sorts the matching factors when both matching factors exceed the predetermined value and provides the output based upon the highest matching factor.

65. An analog or digital element for analyzing biopotential morphologies in myocardial tissue to locate potential ablation sites comprising first means for inputting a template of a cardiac event of known diagnosis in a region of myocardial tissue, second means for inputting a sample of paced biopotentials occurring in the tissue region as a result of a pacing signal from a known pacing site, third means operative in a first mode for electronically comparing the sample to the template using a first comparison technique and generating a first output based upon the first comparison tech- nique, the third means being further selectively operative in a second mode for electronically com¬ paring the sample to the template using a second comparison technique different than the first comparison technique and generating a second output based upon the second comparison technique, the first and second output each comprising a matching factor indicating how alike the input sample is to the input template, and fourth means for comparing at least one of the matching factors to a predetermined value to determine whether the pacing site is a potential ablation site.

66. An element according to claim 65 wherein one of the comparison techniques compares the respective sample to the template by matched filtering.

67. An element according to claim 65 wherein one of the comparison techniques compares the respective sample to the template by cross correlation.

68. An element according to claim 65 wherein one of the comparison techniques compares the respective sample to the template by deriving a norm of the difference.

69. An element according to claim 65 wherein one of the comparison techniques compares the respective sample to the template by using the template to create a matched filtered sample and by analyzing the symmetry of the matched filtered sample.

70. An element according to claim 65 and further including fifth means for comparing the first and second matching factors.

71. An element according to claim 65 wherein the fourth means provides an output when at least one of the matching factors exceeds the predetermined value.

72. An element according to claim 71 wherein the fourth means sorts the matching factors when both matching factors exceed the predetermined value and provides the output based upon the highest matching factor.

73. A system for analyzing biopotential morphologies in myocardial tissue to locate poten¬ tial ablation sites comprising an array of multiple electrodes, an element supporting the multiple elec¬ trode array in operative association with a region of heart tissue, and an analog or digital processing element coupled to the multiple electrode means comprising first means for conditioning the multiple electrodes on the array to sense at each electrode a sample of the biopotentials occurring in the tissue region during a cardiac event of known diagnosis and for creating a template for each electrode based upon the sensed biopotential sample, second means for pacing the heart at a predetermined location while sensing at each electrode a sample of the paced biopotentials occurring as a result pacing, third means for comparing the paced biopotential samples sensed by each multiple elec¬ trode to the templates for each multiple electrode and generate an output comprising a matching factor indicating how alike the paced biopotential samples are to the templates, and fourth means for comparing the matching factor to a predetermined value to determine whether the pacing location is near a potential ablation site.

74. A system according to claim 73 and further including a roving electrode supported for movement relative to the array of multiple electrodes for operative association with selected, different regions of endocardial tissue, and wherein the second means paces using the roving electrode.

75. A system according to claim 73 wherein the second means sequentially conditions different ones of the multiple electrodes on the array to emit a pacing signal and senses at each electrode on the array a sample of the paced biopotentials occurring as a result of the pacing at each different one of the electrodes.

76. A system according to claim 73 wherein the multiple electrodes are in operative association with endocardial tissue to sense electrograms.

77. A system according to claim 73 wherein the multiple electrodes are in operative association with epicardial tissue to sense electrocardiograms.

78. A system according to claim 73 wherein the third means compares the paced biopotential sample to the template by matched filtering.

79. A system according to claim 73 wherein the third means compares the paced biopotential sample to the template by cross corre¬ lation.

80. A system according to claim 73 wherein the third means compares the paced biopotential sample to the template by deriving a norm of the difference.

81. A system according to claim 73 wherein the third means compares the paced biopotential sample to the template by using the template to create a matched filtered paced biopotential sample and by analyzing the symmetry of the matched filtered paced biopotential sample.

82. A system according to claim 73 wherein the sensed biopotential sample of the template comprises an electrogram of a first predetermined duration, and wherein the sensed paced biopotential sample comprises an electrogram of a second prede¬ termined duration not shorter than the first predetermined duration.

83. A system according to claim 82 wherein the first and second predetermined durations are equal.

84. A system according to claim 82 wherein the first and second predetermined durations comprise one or a portion of one heart beat.

85. A system according to claim 73 wherein the fourth means provides an output when the matching factor exceeds the predetermined value.

86. An analog or digital processing element for analyzing biopotential morphologies in myocardial tissue comprising first means for inputting a template of a cardiac event of known diagnosis in a region of myocardial tissue, second means for inputting a first sample of paced biopotentials occurring in the tissue region as a result of pacing using a first technique, third means for inputting a second sample of paced biopotentials occurring in the tissue region as a result of pacing using a second technique different than the first technique, and fourth means for electronically comparing each of the first and second samples individually to the template and generating first and second outputs based upon each comparison.

87. An analog or digital processing element for analyzing biopotential morphologies in myocardial tissue comprising first means for inputting a template of an electrogram having a duration not greater than one heart beat recorded during a cardiac event of known diagnosis in a region of myocardial tissue, second means for inputting a first sample comprising a paced electrogram having a duration not shorter than the template recorded in the tissue region as a result of pacing using a first technique, third means for inputting a second sample comprising a paced electrogram having a duration not shorter than the template recorded in the tissue region as a result of pacing using a second technique different than the first technique, and fourth means for electronically comparing each of the first and second samples individually to the template and generating first and second outputs based upon each comparison.

88. An analog or digital processing element for analyzing biopotential morphologies in myocardial tissue comprising first means for inputting a template of an electrogram having a duration not greater than one heart beat recorded during a cardiac event of known diagnosis in a region of myocardial tissue, second means for inputting a first sample comprising a paced electrogram having a duration not shorter than the template recorded in the tissue region as a result of pacing using a first technique, third means for inputting a second sample comprising a paced electrogram having a duration not shorter than the template recorded in the tissue region as a result of pacing using a second technique different than the first technique, one of the first and second techniques comprising entrainment or reset pacing, and fourth means for electronically comparing each of the first and second samples individually to the template and generating first and second outputs based upon each comparison.

89. An analog or digital processing element for analyzing biopotential morphologies in myocardial tissue comprising first means for inputting a template of an electrogram having a duration not greater than one heart beat recorded during a cardiac event of known diagnosis in a region of myocardial tissue, second means for inputting a first sample comprising a paced electrogram having a duration not shorter than the template recorded in the tissue region as a result of pacing using a first technique, third means for inputting a second sample comprising a paced electrogram having a duration not shorter than the template recorded in the tissue region as a result of pacing using a second technique different than the first technique, one of the first and second techniques comprising entrainment or reset pacing and the other one of the first and second techniques comprising pace mapping, and fourth means for electronically comparing each of the first and second samples individually to the template and generating first and second outputs based upon each comparison.

90. An element according to claim 86 or 87 or 88 or 89 wherein at least one of the outputs comprises a matching coefficient indicating how alike the respective sample is to the template.

91. An element according to claim 86 or 87 or 88 or 89 wherein the fourth means compares at least one of the samples to the template by matched filtering.

92. An element according to claim 86 or 87 or 88 or 89 wherein the fourth means compares at least one of the samples to the template by cross correlation.

93. An element according to claim 86 or 87 or 88 or 89 wherein the fourth means compares at least one of the samples to the template by deriving a norm of the difference.

94. An element according to claim 86 or 87 or 88 or 89 wherein the fourth means compares at least one of the samples to the template by using the template to create a matched filtered sample and by analyzing the symmetry of the matched filtered sample.

95. An element according to claim 86 or 87 or 88 or 89 wherein the fourth means compares one of the samples to the template using a first prescribed methodology to generate the first output and compares the other sample to the template using a second prescribed methodology different than the first methodology to generate the second output.

96. An element according to claim 95 and further including fifth means for comparing the first and second outputs.

97. An element according to claim 86 or 87 or 88 or 89 and further including fifth means for comparing the first and second outputs.

98. A system for analyzing biopotential morphologies in myocardial tissue comprising electrode means for sensing biopotentials in myocardial tissue, and a processing element coupled to the electrode means comprising first means for conditioning the elec¬ trode means to sense a sample of the biopotentials occurring in the tissue region during a cardiac event of known diagnosis and for creating a template based upon the sensed biopotential sample, second means operative in a first mode for conditioning the electrode means to pace using a first technique and to sense using the electrode means a first sample of the paced biopotentials occurring in the tissue region as a result of the first pacing technique, the second means being further selectively operative in a second mode for conditioning the electrode means to pace using a second technique different than the first technique and to sense using the electrode means a second sample of the paced biopotentials occurring in the tissue region as a result of the second pacing technique, and third means for electronically com¬ paring each of the first and second samples individually to the template and generating first and second outputs based upon each comparison.

99. A system according to claim 98 and further including an element supporting the electrode means for operative association with a region of heart tissue.

100. An analog or digital element for analyzing biopotential morphologies in myocardial tissue comprising first means for inputting a template of a cardiac event of known diagnosis in a region of myocardial tissue, second means for inputting a sample of paced biopotentials occurring in the tissue region as a result of a pacing signal, third means operative in a first mode for electronically comparing the sample to the template using a first comparison technique and generating a first output based upon the first comparison technique, the third means being further selectively operative in a second mode for electronically com- paring the sample to the template using a second comparison technique different than the first comparison technique and generating a second output based upon the second comparison technique.

101. An analog or digital element for analyzing biopotential morphologies in myocardial tissue comprising first means for inputting a template of an electrogram having a duration not greater than one heart beat recorded during a cardiac event of known diagnosis in a region of myocardial tissue, second means for inputting a sample of a paced electrogram having a duration not shorter than the template recorded in the tissue region, third means operative in a first mode for electronically comparing the sample to the template using a first comparison technique and generating a first output based upon the first comparison technique, the third means being further selectively operative in a second mode for electronically com¬ paring the sample to the template using a second comparison technique different than the first comparison technique and generating a second output based upon the second comparison technique.

102. An analog or digital element for analyzing biopotential morphologies in myocardial tissue comprising first means for inputting a template of an electrogram having a duration not greater than one heart beat recorded during a cardiac event of known diagnosis in a region of myocardial tissue, second means for inputting a sample of a paced electrogram having a duration not shorter than the template recorded in the tissue region during entrainment or reset pacing, third means operative in a first mode for electronically comparing the sample to the template using a first comparison technique and generating a first output based upon the first comparison technique, the third means being further selectively operative in a second mode for electronically com¬ paring the sample to the template using a second comparison technique different than the first comparison technique and generating a second output based upon the second comparison technique.

103. A system for analyzing biopotential morphologies in myocardial tissue comprising electrode means for sensing biopotentials in myocardial tissue, and a processing element coupled to the electrode means comprising first means for conditioning the elec¬ trode means to sense a sample of the biopotentials occurring in the tissue region during a cardiac event of known diagnosis and for creating a template based upon the sensed biopotential sample, second means operative for conditioning the electrode means to emit a pacing signal and to sense using the electrode means a sample of the paced biopotentials occurring in the tissue region as a result of the pacing signal, and third means operative in a first mode for electronically comparing the paced biopotential sample to the template using a first comparison technique and generating a first output based upon the first comparison technique, the third means being further selectively operative in a second mode for electronically comparing the paced biopotential sample to the template using a second comparison technique different than the first comparison technique and generating a second output based upon the second comparison technique.

104. A system according to claim 103 and further including an element supporting the electrode means for operative association with a region of heart tissue.

105. A system for analyzing biopotential morphologies in myocardial tissue comprising electrode means for sensing biopotentials in myocardial tissue, and a processing element coupled to the electrode means comprising first means for conditioning the elec¬ trode means to sense a sample of the biopotentials occurring in the tissue region during a cardiac event of known diagnosis and for creating a template based upon the sensed biopotential sample, second means operative in a first pacing mode for conditioning the electrode means to pace using a first technique and to sense using the electrode means a first sample of the paced biopotentials occurring in the tissue region as a result of the first pacing technique, the second means being further selectively operative in a second pacing mode for conditioning the electrode means to pace using a second technique different than the first pacing technique and to sense using the electrode means a second sample of the paced biopotentials occurring in the tissue region as a result of the second pacing technique, and third means operative in a first comparison mode for electronically comparing each of the first and second paced biopotential samples individually to the template using a first comparison technique and generating first and second outputs based upon each comparison using the first comparison technique, the third means being further selectively operative in a second comparison mode for electronically comparing each of the first and second paced biopotential samples individually to the template using a second comparison technique different than the first comparison technique and generating third and fourth outputs based upon each comparison using the second comparison technique.

106. A system according to claim 105 and further including an element supporting the electrode means for operative association with a region of heart tissue.