CA2228579A1 - Slow gain control - Google Patents

Abstract

A system and method automatically controls a gain of a cardioverter/defibrillator which receives electrical activity of the heart and delivers shock pulses in response thereto. An amplifier amplifies the electrical activity according to a variable gain. A detection circuit detects depolarizations in the amplified electrical activity and provides a detect signal representing a cardiac event indicative of a depolarization when the amplified electrical activity exceeds a sensitivity threshold. A storage device stores peak history information representative of peak values of the amplified electrical activity of a first selected number (N) of cardiac events. Slow gain control circuitry adjusts the variable gain in discrete steps based on the stored peak history information.

Description

W O 97/06852 PCT~US96113205 SI,OW GAIN CONTROL

s Field of the Invention The present invention relates generally to implantable medical devices, and more particularly, to systems such as ~lltc)m~tic gain control systems for ~l~tQm~tic~lly adjusting the sensing threshold in cardiac rhythm management devices, such as p~c~m~k~rs, cardioverter/defibrillators, and 10 cardioverter/defibrillators with pacing capability.
R~ ~k~round of thelnvention Cardiac rhythm management devices such as pa~em~kPrs, cardioverter/defibrillators, and cardioverter/defibrillators with pacing capability typically include a system for detectin~ dangerous cardiac alll.yl~.,..ia conditions 15 in the heart, such as bradycardia, tachycardia, and fibrillation by ~n~ the time interval between con~ecutive cardiac depolarizations. Cardiac rhythm management devices receive a sensed cardiac signal comprising electrical activity of the heart and detect cardiac depolarizations in the electrical activity when an amplitude of the electrical activity exceefl~ a predetermine~ amplitude 20 level or "sensing threshold." The sensing threshold may be fixed, or may vary over time.
A fixed sensing threshold is not ~p~lu~liate for detecting certain ~hylhlllias, such as polymorphic tachycardia and fibrill~tion, wherein extreme variations occur in the amplitude of the electrical activity during the arrhythmia.
25 The problem of tracking variations in the ~mplitll(ie of the electrical activity is further complicated when the cardiac rhythm management device delivers pace pulses to the heart, which cause invoked responses which are quite high in ~mp!itllde as conlp~;d to normal cardiac depolarizations.
One approach to com~.ell~al~ for problems associated with a fixed 30 sensing threshold is to prograrn the sensing threshold at a value determined by the atterl~lin~ physician after careful study of the variety of amplitudes in cardiac signal activity experienced by a patient. In other words, a sensing threshold is CA 02228~79 1998-02-04 programmed into the cardiac rhythm m~n~gem-ont device, and any cardiac signal amplitude larger than the programmed sensing threshold is considered a cardiac depolarization. If, however, the prog.,."~ ed sensing threshold is set too high and the cardiac signal amplitude decreases .~i~nific~ntly, as is often the case in S fibrill~tic n, the cardiac rhythm management device may not sense the allhyll.. . ia. If the programmed sensing threshold is set too low, the device may over-sense. For ç~mple7 a system designed to detect ventricular depol~ri7~til-n~(R-waves) may e.lulleously detect atrial depolarizations (P-waves) or ventricular recovery (T-waves). B~n-lp~c~ filtering can be used to partially elimin~t~
10 erroneous detection of the P-waves and T-waves in a R-wave detection system.
If, however, the band of frequencies passed by the b~n-lp~cc filt~ring is too narrow, certain fibrillation signals may not be detecte~l Another approach to compt;llsaLe for the above problems is to set the sensing threshold proportional to the amplitude of the sensed cardiac signal15 each time a cardiac depolarization is sensed. The sensing threshold is then allowed to decrease over time between consecutively sensed cardiac depolarizations so that if the sensed cardiac signal ~mplitll(le decl~ases significantly, the cardiac rhythm management device is still able to detect the lower level amplitude of the cardiac signal. Adjusting the sensing threshold to 20 an ~ro~liate level with this approach becomes difficult if the patient requires pacing due to a bradycardia condition. For example, in a system that senses R-waves acco~ g to this approach, the sensing threshold may be adjusted to one-half of the R-wave arnplitude when an R-wave is sensed. However, the invoked response due to a first pacing pulse can cause the sensing threshold to be 25 set so high that a second spontaneous R-wave is not sensed. Because the system does not sense the second spontaneous R-wave, a second pacing pulse is delivered to the patient in~pr,~pliately.
One solution to the above problem is found in the Kelly et al.
U.S. Patent No. 5,269,300 5~ ned to Cardiac P~c~m~k~rs, Inc., the assignee of 30 the present application. The Kelly et al. patent rli~closes an impl~nt~hle cardioverter/defibrillator with pacing capability wh~ the sensing threshold is W O 97/06852 PCT~US96/13205 tom~ti~ y adjusted to a value s~lol)ol lional to the amplitude of the sensed cardiac signal. The sensing threshold continuously decreases between sensed cardiac depolarizations to ensure that a lower level cardiac signal will be detected. However, after a pacing pulse is delivered by the Kelly et al. device,5 the s~ ing threshold is set to a fixed value, and held at the fixed value for a pre~let~rmin~od period of time, so that the sensing threshold is not affected by the cardiac response invoked by the pacing pulse. After a predet~rminçd period of time, the sensing threshold is decreased, just as after a s~o~ eou~ cardiac depolarization.
In the Keimel et al. U.S. Patent No. 5,117,824, an R-wave detector automatically adjusts the detecting threshold in response to the R-waveamplitude. The adju~tment of the threshold is disabled for a pred~tf~nnint?d period following the deliver,v of each pacing pulse. Thclc~rh,l, the sensing threshold is returned to a lower threshold level to allow detection of lower level 15 R-waves indicative of tacl.y,llylhlia conditions.
In the Henry et al. U.S. Patent No. 5,339,820, a sensitivity control is used for controlling a sensing threshold in a cardiac control device such as a pacemaker, cardioversion and/or cardiac defibrillation device. Initially, a sensing threshold is set to a low value. When the cardiac signal is detectP-l, the amplitude of the R-wave is measured and the sensing threshold is computed as a function of the amplitude of the R-wave. After a refractory period, the sensing threshold is preferably set to 75% of the amplitude of the R-wave. The sensing threshold is then decreased in uniform steps. The uniform steps may be fixe decrements or percentage reductions.
The Gravis et al. U.S. Patent No. 4,940,054 discloses a cardioversion device having three sensitivities. A first, mediuln sensitivity isused for the detection of sinus rhythrn and ventricular tachycardia. A second, higher sensitivity is designed for dirr~le." i~ing ventricular fibrillation fromasystole. A third, lower sensitivity is used to dirr~ iate between R-waves and high ~mr~lih~-1e current of injury T-waves which occur after shocking. One of these three sensitivities is selected as a function of the status of the device, such as during a period of suspected t~chycaldia or a post shock period, and the selected sensitivity must be ,,, ;,,ls-;,,P~1 at least until the next cycle.
The Dissing et al. U.S. Patent No. 5,370,124 discloses a cardiac rhythm management device having cilCuilly for automatically adapting the S detection SG11Si~iVilY to the cardiac signal. The detection sell~ilivi~y is adjusted by either amplifying the electriç~l signal supplied to the threshold detector with a variable gain given a p. ~ ly prescribed threshold or by varying the threshold itself. In either case, the GrrG~ilivG threshold is based on an average value formed over a time interval coll~i,pollding to the duration of a few breaths.
10 A switching hy~L~,- ;,is is generated having a lower limit value and an upper limit value, where the threshold is reset only when the average value falls below the lower limit value or exceeds the upper limit value. The limit values of the :iwilchillg hy~l~ ,esis are varied with the variation of the threshold, but the relationship of the limit values to the threshold remain unvaried. In one 15 embo-lim~nt of the Dissing device, when the threshold is set below a miniml-mvalue, a beat-to-beat variance of signal heights of suçceccive input electrical signals are used for fonning an average value. The sensing threshold is raised by a preclet~rmin~cl amount if the variance exçee-lc the predet~rmin~d variance value.
The Carroll et al. U.S. Patent No. 4,972,835 discloses an ii-~pli~ hle cardiac defibrillator which includes switched C,Z~p~lritor ch~;uill,y for amplifying the cardiac electrical signal with non-binary gain çh~ngin~ steps.
Three stages of gain are used to increase the gain approximately 1.5 each illGl GlllGllt.
The Baker et al. U.S. PatentNo. 5,103,819 discloses a state mzlrhinP~ for automatically controlling gain of the sensing function in an implantable cardiac stim~ tor. The rate of gain adjllctment is dependent on the present sensed conditions and on the prior state of the heart. Different rates of adjnctm~nt are selected under varying conditions so that the gain of the sense amplifier is adjusted without significant overshoot. Multiple effective time CA 02228~79 1998-02-04 constants are used for different conditions by basing the rate of adjustment of the sense amplifier gain on the path kaversed in the state m~.hine The Baker, Jr. et al. U.S. Patent 4,888,004 describes a R-wave sensor having a band pass amplifier and a comparator circuit to determine when the band passed signal exceeds a predetermined, fixed sensing threshold to determine a sensed event. A gain is decreased if a AGC target is crossed.
Therefore, considerable effort has been expended in providing for automatically adjustable sensing thresholds through adjusting the threshold level itself or with automatic gain circuitry in implantable cardiac rhythm management10 devices for the purpose of enhancing the capability of the device to sense arrhythrnia conditions for which therapy is to be applied.
Summary of the Invention The present invention provides a method and system for automatically conkolling a gain of a cardioverter/defibrillator which receives 15 eleckical activity of the heart and provides shock pulses in response to the received electrical activity. An amplifier amplifies the electrical activity of the heart according to a variable gain. A cardiac depolarization detector detects depolarizations in the amplified electrical activity of the heart and provides adetect signal representing a cardiac event indicative of a depolarization when the 20 amplified eleckical activity exceeds a sensitivity threshold. A storage device stores peak history information representative of peak values of the arnplified eleckical activity of a first selected number (N) of cardiac events. A gain conkoller adjusts the variable gain in discrete steps based on the stored peak history information.
The gain controller preferably increases the variable gain by at least one discrete step if a second selected number (M) of peak values of the N
cardiac events are below a selected low threshold and decreases the variable gain by at least one discrete step if M peak values of the N cardiac events are above a selected high threshold. The value of M is preferably at least 3 and the value of 30 N is preferably at least 4. The peak history information from the previous cardiac event is updated in the storage device preferably at the beginning of a ..

Sa new refractory period caused by a cardiac event. The storage device preferably includes a first group of storage locations which store peak inforrnation indicating if the peak values are below the selected low threshold and a second Fn S~ET

group of storage locations which store peak information inrlic~tin~ if the peak values are above the selectecl high threshold.
In a ~c~rtll~ d embodiment of the present invention, the gain controller i.lc.eases the variable gain if a stored peak value of a last cardiac event S and M - 1 peak values of the last N - 1 cardiac events previous to the last cardiac event are below the selected low threshold and decreases the variable gain if the stored peak value of the last cardiac event and the M - l peak values of the last N
- 1 cardiac events previous to the last cardiac event are above the selected high threshold.
In a pl~ef~ ;d embodiment of the present invention, the gain controller inrhl~les a storage device capable of storing peak historv information l~les~ e of peak values of amplified electrical activity of a third selected number of cardiac events. In this embodiment, far field sense Cil~;uilly responds to the stored peak history h~ro~ lion to indicate a decrease in the variable gain 15 if the peak values of the amplified electrical activity of the third selected nurnber of cardiac events ~ltern~te between clipped peak values and non-clipped peak values. The peak value is d~ to be clipped if the peak value is at a maximum peak value. In the p.cr~ d embodiment of the present invention, the gain controller is imI~lem~nte-l in digital cilcuilly, which permits a simple 20 comparison to a m~xj--l.-. digital value to cletermine whether or not a peak value has been clipped. The third selected number can be equal to the first selected number. In the embodiment of the present invention where M is equal to 3 and N is equal to 4, the third selected nurnber is preferably set to a value of 6 which uileS two extra storage locations in the storage device.
In a ~ler~ d embodiment of the present invention, the gain controller includes cir~uil.r responsive to the detect signal to set the gain to a selected relatively high sensitivity if a cardiac event is not detected for a selected time period. For ex~mple7 a suitable selected time period is ~ oxilnately 1.5 seconds which is equivalent to approximately 40 heart beats per minute. The selecte~l relatively high sensitivity is preferably at least one discrete gain step from a ~ tx;ll~ sensitivity~ The gain controller also preferably includes CA 02228~79 1998-02-04 circuitry to set the gain to the selected relatively high sensitivity when the cardioverter/defibrillator delivers a shock pulse to the heart. In the cardioverter/defibrillator with pacing capability embodiment of the present invention, the gain controller also preferably includes circuitry to set the gain to the selected relatively high sensitivity when the cardioverter/defibrillator delivers a pacing pulse to the heart.
The gain controller preferably includes circuitry to decrement the gain from the selected relatively high sensitivity by a selected nurnber of discrete gain steps if the setting of the gain to the selected relatively high sensitivity 10 creates a clipped peak value of the amplified electrical activity on the following detected cardiac event. The circuitry preferably further decrements the gain by at least one discrete gain step if the peak value of the amplified electrical activity is still clipped on the second detected cardiac event following the setting of the gain to the selected relatively high sensitivity. In addition, the circuitry 15 preferably decrements the gain from the selected relatively high sensitivity by a selected nurnber of discrete gain steps if the setting of the gain to the selected relatively high sensitivity does not create a clipped peak value of the arnplified electrical activity on the following detected cardiac event and does create a clipped peak value of the arnplified electrical activity on the second detected 20 cardiac event following the setting of the gain to the selected relatively high sensitivity.
Brief Description of the Drawin~s Figure 1 is a block diagram of a dual chamber cardioverter/defibrillator .
Figure 2 is a logical block diagrarn of an AGC filter and digitizing circuit.
Figure 3 is a timing diagram illustrating the sensed refractory used in the cardioverter/defibrillator of Figure 1.
Figure 4 is a tirning diagram illustrating the paced/shock 30 refractory used in the cardioverter/defibrillator of Figure 1.
~,,q~U~

r CA 02228~79 1998-02-04 .
8 ~ .
Figure 5 is a logical block diagram of a gain conkol circuit.
Figure 6 is a timing diagram illustrating a piecewise linear approximation of an exponential decay of the variable sensing threshold.
Figure 7 is a template generation circuit according to the present 5 invention, which achieves the piecewise linear approximation of an exponential decay illustrated in Figure 6.
Figure 8 is a timing diagraIn illustrating the operation of the slow gain control circuit of Figure 5, in combination with the fast templating generation circuit of Figure 7 in adjusting the gain and the sensing threshold of 10 the cardioverter/defibrillator.
Description of the Preferred Embodiments In the following detailed description of the preferred embodiments, reference is made to the accompanying drawings which form a part hereof, and in which is shown by way of illustration specific embodiments l S in which the invention may be practiced. It is to be understood that other embodiments may be utilized and structural or logical changes may be made without departing from the scope of the present invention.
~)ual Chamber Cardioverter/Defibrillator with Pacing Capabilitv A dual chamber cardioverter/defibrillator 20 with pacing 20 capability is illustrated in block diagrarn form in Figure 1.
Cardioverter/defibrillator 20 operates as a pulse generator device portion of a cardiac rhythm management system which also includes leads or electrodes (not shown) disposed in the ventricular chamber of the heart to sense electrical activity representative of a R-wave portion of the PQRST complex of a surface 25 EGM indicating depolarizations in the ventricle. Cardioverter/defibrillator 20 includes inputloutput terminals 22 which are connectable to the ventricular leads to receive the ventricular electrical activity of the heart sensed by the ventricular - CA 02228~79 1998-02-04 leads. A pace pulse circuit 24 provides pacing pulses such as bradycardia and antitachycardia pacing pulses to input/output terminals 22 to be provided to theventricular chamber of the heart via the ventricular leads to stimulate excitable myocardial tissue to treat arrhythmia conditions such as bradycardia and some 5 tachycardia. A shock pulse circuit 26 provides shock pulses to input/output terminals 22 to be provided to the ventricular charnber of the heart via the ventricular leads to shock excitable myocardial tissue to treat tachyrhythmia conditions. The tachyrhythmia conditions may include either ventricle fibrillation or ventricle tachycardia.
A filter and after potential removal circuit 28 filters the ventricular electrical activity received by input/output terminals 22 and the pacing pulses provided from pacing pulse circuit 24. In addition, ~1lter and after potential removal circuit 28 removes after potential created by a pacing pulse from pacing pulse circuit 24 or a shock pulse delivered by shock pulse circuit 26.
An automatic gain control (AGC)/filter and digitizing circuit 30 amplifies the filtered ventricular electrical activity provided from the filter and after potential removal circuit 28. AGC/filter and digitizing circuit 30 includes circuitry for digitizing the filtered venkicular electrical activity. A gain control circuit 32 automatically adjusts the gain of AGC/filter and digitizing circuit 3 O.
An R-wave detection circuit 34 is coupled to AGC/filter and digitizing circuit 30 to detect depolarizations in the amplified ventricular electrical activity representative of R-wave depolarizations when the amplified ventricular electrical activity exceeds a selected amplified level known as the "sensitivity threshold" or the "sensing threshold" and refractory is 25 inactive. A template generation circuit 36 automatically selects and adjusts the sensing threshold. R-wave detection circuit 34 provides a R-wave CA 02228~79 1998-02-04 depolarization signal, indicative of the R-wave depolarizations, to a microprocessor and memory 38.
The cardiac rhythm management system also includes leads or electrodes (not shown) disposed in the atrial chamber of the heart to sense 5 electrical activity representative of a P-wave portion of the PQRST complex of a surface EGM indicating depolarizations in the atrium. Cardioverter/defibrillator20 correspondingly also includes input/output terminals 42 which are connectable to the atrial leads to receive the atrial electrical activity of the heart sensed by the atrial leads. A pace pulse circuit 44 provides pacing pulses such as 10 bradycardia pacing pulses to input/output terminals 42 to be provided to the atrial chamber of the heart via the atrial leads to stimulate excitable myocardial tissue to treat arrhythmia conditions such as bradycardia or atrial tachycardia. A
filter and after potential removal circuit 48 operates similar to filter and after potential removal circuit 28 to filter the atrial electrical activity received by 15 input/output terminals 42 and the pacing pulses provided from pacing pulse circuit 44. In addition, filter and after potential removal circuit 48 removes after potential created by a pacing pulse from pacing pulse circuit 44.
An automatic gain control (AGC)/filter and digitizing circuit 50 amplifies the filtered atrial electrical activity provided from the filter and after 20 potential removal circuit 48. AGC/filter and digitizing circuit 50 includes circuitry for digitizing the filtered atrial electrical activity. A gain control circuit 52 automatically adjusts the gain of AGC/filter and digitizing circuit 50. An P-wave detection circuit 54 is coupled to AGC/filter and digitizing circuit 50 to detect depolarizations in the amplified atrial electrical activity representative of 25 P-wave depolarizations when the amplified atrial electrical activity exceeds a selected amplified level known as the "sensitivity threshold" or the "sensing threshold" and the refractory is inactive. A template generation circuit 56 automatically selects and adjusts the sensing threshold. P-wave detection circuit 54 provides a P-wave depolarization signal, indicative of the P-wave 30 depolarizations, to microprocessor and memory 38.
~ S~

CA 02228~79 1998-02-04 W O 97/06852 PCTnUS96/13205 Microprocessor and memory 38 analyzes the ~etec~l P-waves in~lir~ted in the P-wave depolarization signal from P-wave detection circuit 54 along with the R-wave depolarization signal provided from R-wave detection circuit 34 for the detection of ~llly~ llia con-lition~ based on known al~olilhllls.
5 For example, microprocessor and memory 38 can be used to analyze the rate, regularity, and onset of variations in the rate of the reoccurrence of the ~letected P-wave and/or R-wave, the morphology of the ~ietected P-wave andl/or R-wave, or the direction of propagation of the depolarization represented by the detected P-wave and/or R-wave in the hea~t. In addition, microprocessor and memory 38 10 stores depolarization data and uses known techniques for analysis of the detectecl R-waves to control pace pulse circuit 24 and shock pulse circuit 26 for deliveryof pace pulses and shock pulses to the ventricle and for analysis of detected P-waves to control pace pulse circuit 44 for proper delivery of pace pulses to theatrium. In addition, microprocessor and memory 38 controls a state m~hine 39 15 which places various circuits of cardioverter/defibrill~t~lr 20 in desired logical states based on various conditions such as when a pace pulse or shock pulse occurs or on O~ g conditions of the cardioverter/defibrillator such as bradycardia pacing, tachyrhythmia sensing, and normal sinus sen.~ing The dual c h~mh~r cardioverter/defibrillator 20 with pacing 20 capability illustrated in Figure 1 includes pacing and shocking capabilities for the ventricle and pacing capability for the atrium. Nevertheless, the present invention can be embodied in a single chamber cardiac rhythm management device having a single one of these capabilities. For example, the present invention can be embodied in a ventricle defibrillator device for providing shock 25 pulses to the ventricle only.
In some embo-lim~nt~ of cardioverter/defibrillator 20, input/output t~nin~lc 22 and 42 are each implem~nt~-l to be connectable to a COll~ sponding single set of electrodes (not shown) used for pacing, shock delivery, and sensing. In other embodimpnt~ of cardioverter/defibrillator 20, the 30 input/output t~rmin~l~ are implemented to be connectable to separate sets of electrodes for pulse delivery and sensing. In some embodiments, the CA 02228~79 1998-02-04 input/output te.l~ als are impletnented to be connectable to separate electrodesfor pacing and shock delivery. In all of these embo~1im~nt~, the eleckodes of a cardiac rhythm management system are typically implemented as unipolar or bipolar eleckrodes.
A unipolar electrode configuration has one pole or electrode (i.e., negative pole or cathode electrode) located on or within the heart, and the other pole or eleckode (i.e., positive pole or anode electrode) remotely located from the heart. With endocardial leads, for example, the cathode is located at the distal end of a lead and typically in direct contact with the e~loç~rdial tissue to 10 be stim~ tetl~ thus forming a "tip" eleckrode. Conversely, the anode is remotely located from the heart, such as comprising a portion of the metallic enclosure which surrounds the implanted device, thus forming a "can" electrode and is often referred to as the "indifferent" electrode.
A bipolar eleckode configuration has both poles or eleckrodes 15 typically located within the atrial or ventricular chamber of the heart. Withendocardial leads, for example, the cathode is located at the distal end of the lead, referred to as the "tip" eleckode. In the bipolar configuration, the anode is usually located approximate to the "tip" eleckode spaced apart by 0.5 to 2.5 cm., and typically forming a ring-like structure, referred to as the "ring" electrode.
With respect to s~n~ing, it is well known that bipolar and unipolar electrode configllr~tione do not yield equivalent cardiac EGMs. Each configuration has advantages and disadvantages, for example, with a unipolar-sensing configuration, only the electrical events ~ qc~nt to the "tip" electrodecontrol the unipolar EGM, while the remote "indifferent" electrode contributes 25 negligible voltage due to its location being extracardiac.
With a bipolar-sensing configuration, the magnitude of the cardiac signal is similar for both the "ring" and the "tip" electrodes, but the reslllting EGM is highly dependent upon the u~ ;on of the electrodes within the heart. Optimal sensing will occur, for example, when the sensing vector 30 defined by the sensing electrodes is parallel with the dipole defined by the depolarization signal. Since bipolar electrodes are more closely spaced than their unipolar cou~ , the depolarization signal will be shorter in duration than that produced from a unipolar configuration. Due to a more restrictive leadfield or ~ n~, bipolar sensing offers improved rejection of ele~ gllPtic and skeletal muscle artifacts, and thus provides a better signal-to-noise ratio than - 5 unipolar sensing.
A.GC/Filter ~nti D;~iti7ir~ Circuit A logical block diagram representative of AGC/filter and iti7in~ circuit 30 or 50 is illustrated in Figure 2. A pro~ ble gain filter 60 filters the electrical activity provided from the filter and after potential removal circuit 28 or 48 of ~igure 1. When cardioverter/defibrillator 20 of Figure 1 is implPm~nte~l to be connectable to bipolar electrodes, progr~mm~hle gain filter 60 comrri~es an analog differential sense amplifier to sense and arnplify the di~lel~ce b~;Lw~.. first and second bipolar electrodes.
Programmable gain filter 60 has a programmable gain to initially amplify the 15 incoming electrical activity.
An analog to digital (A/D) co~ ,. L~,l 62 receives the filtered and amplified electrical activity from programmable gain filter 60 and converts the analog electrical activity to ~1igiti7:Pd cardiac data, which is stored in a succe~sive approximation register (SAR) 64. A/D converter 62 operates by co.~-p~. ;..g a 20 sample of "unknown" analog electrical activity from programmable gain filter 60 against a group of weighted values provided from SAR 64 on lines 66. A/D
co~ 62 compares the weighted values on lines 66 in descPntling order, starting with the largest weighted value. A weighted value is not added to the summ~-l digital data stored in SAR 64 if the weighted value, when added to the 25 previous s~mmçrl weighted values, produces a sum larger than the sampled "unknown" analog electrical activity. The summed digital data is updated in SAR 64 and a new weighted value is compaled on each active edge of a SAR
clock on a line 68.
At the end of the successive approximation when balance is 30 achieved, the sum of the weighted values stored as the summed digital data inSAR 64 l~cse~ the approxim~tecl value of the sampled "unknown"analog -CA 02228~79 1998-02-04 electrical activity. SAR 64 provides the stored digital cardiac data to an absolute value circuit 70. Absolute value circuit 70 provides the absolute value of the amplitude of the digital cardiac data on a line 72 to be provided to gain control circuit 32/52 and template generation circuit 36/56. Successive approximation 5 A/D conversion as performed by A/D converter 62 and SAR 64 is very fast to permit adequate tracking of the incoming analog cardiac signal. The gain of programmable gain filter 60 is raised or lowered in discrete gain steps based onoutputs from gain control circuit 32/52.
Separate Gain Control and Threshold Templating Gain control circuit 32/5Z and template generation circuit 36/56 operate with the AGC/filter and digitizing circuit 30/50 to implement two independent AGC digital loops. Gain control circuit 32/52 provides slow gain control to AGC/filter and digitizing circuit 30/50 to keep sénsed depolarizations representative of cardiac events in approximately the upper one third of the 15 dynamic range of A/D converter 62. Template generation circuit 36/56 providesa fast responding variable sensing threshold to the detection circuit 34/54 for actual sensing of R-wave or P-wave depolarizations representative of cardiac events.
Gain control circuit 32/52, as described in more detail below with 20 reference to Figure 5, stores peak history information representative of peakvalues of the amplified electrical activity of a selected number (N) of cardiac events. Gain control circuit 32/52 adjusts the variable gain of AGC/filter and digitizing circuit 30/50 in discrete steps based on the stored peak history information. The stored peak history information is compared against 25 predefined levels and appropriate gain changes are initiated based on a second selected number (M) of peak values of the N cardiac events being outside of a selected range.
Template generation circuit 36/56, as described in more detail below with reference to Figure 7, provides a time varying sensing threshold to 30 detect circuit 34/54 for comparison to the digitized cardiac data provided on line 72 from AGC/filter and 11igiti~:ing circuit 30/50. Detection circuit 34/54 ~ S~

CA 02228~79 1998-02-04 W O 97/06852 PCT~US96/13205 provides a detection signal in~iic:lting R-wave or P-wave depol~ri7~tion~
represçnt~tive of cardiac events when the value of the incoming digital cardiac data is greater than the sensing threshold level provided that the refractory windows are illactive. Template generation circuit 36/56 includes cil.;uillr for~ S selecting and adjusting the variable sensing threshold to a level p~ Jol Lional to the amplitude of the digital cardiac data on line 72. Typically, tell,plale generation circuit 36/56 responds very quickly to change the sensing threshold to the peak value of the digital cardiac data on line 72. The variable sensing threshold is held at the peak value for a selected period of time after which the variable sensing threshold drops to a percentage of the peak value. The variablesensing threshold is then allowed to slowly decay from this percentage of peak value in discrete steps until the variable sensing threshold is at a low threshold value. Template generation circuit 36/56 preferably employs integer math to achieve a piecewise linear ~ xi-llation of a geometric progression such as an expon~nti~l decay curve with minim~l error between piecewise steps.
Refractory Periods Cardioverter/defibrillator 20 utilizes ventricular and atrial refractory periods to ~l~ t~ . ., .il ,e which sensed events are R-waves or P-waves respectively. The active sensed refractory periods are illustrated in timing diagram form on line 73 at 74 in Figure 3. Any sensed event that occurs when the sensed refractory period is inactive is considered to be a R-wave or P-wave.Any events sensed during the active sensed refractory period are ignored and do not affect the ventricular or atrial cycle length mea~u~ nent. Typical sensed events occ~ ng on the lead are represented on line 75 at 76. As illustrated, thestart of the active ventricular or atrial refractory period is synchronized with the start of the cardiac cycle. An absolute refractory interval is indicated on line 77 at 78. The absolute refractory interval starts at the beginning of the cardiac cycle .cimlllt~n.oous with the start of the active sensed refractory period. The absolute refractory interval disables all sensing. The operation of template generation circuit 36/56 based on the absolute refractory interval is further described below under the Threshold Templating for a Fast Digital AGC Circuit hezl~1ing CA 02228~79 1998-02-04 During pacing or shock delivery from cardioverter/defibrillator 20 a paced/shock refractory period, as indicated on line 79 at 81 in Figure 4, is utilized instead of the sensed refractory period. Similar to the sensed refractory period, any sensed event that occurs when the paced/shock refractory period is 5 inactive is considered to be a R-wave or P-wave. Typical pace pulses on the lead are represented for illustrative purposes on line 83 at 85. A typical shock pulse is not shown. The paced/shock refractory period is started with the deliverv of the pace or shock pulse. Absolute refractory intervals are not utilized during pacing or shocking conditions. The time duration of the paced refractory period 10 is preferably programmable, while the time duration of the shock refractory period is typically not programmable. The paced refractory period can be selected by the physician and programmed into cardioverter/defibrillator 20 when the cardioverter/defibrillator is operating in a pacing mode. The operationof template generation circuit 36/56 based on the paced/shock refractory period 15 is further described below under the Threshold Templating for a Fast Digital AGC Circuit heading.
Slow Gain Control Circuit Gain control circuit 32, or alternately gain control circuit 52, is representatively illustrated in Figure 5 in logical block diagram form. A
20 comparator 80 receives the digital cardiac data on line 72 and compares the peak value of the digital data representing the current cardiac event to a selected low threshold and a selected high threshold. For example, in one preferred embodiment where the maximum value of the peak value of the digital cardiac data is 7F hex, the selected low threshold is 52 hex and the selected high 25 threshold is 7E hex. A first storage register 82 includes a first group of storage locations which store peak history information provided by comparator 80 on a line 84 indicative of whether the peak values are below the selected low threshold (52 hex in the example embodiment). A second storage register 86 includes a second group of storage locations which store peak history 30 information provided by comparator 80 on a line 88 indicative of whether the ) S~

CA 02228~79 1998-02-04 peak values are above the selected high threshold (7E hex in the exarnple embodiment).
An M/N circuit 90 receives peak history information from storage register 82 and determines if M peak values of N cardiac events are below the S selected low threshold (52 hex). M/N circuit 90 provides an increment signal on a line 92 to a gain control clock circuit 94. M/N circuit 90 activates the increment signal on line 92 when M out of N peak values are below the selected low threshold (52 hex) to indicate that the gain of AGC/filter and digitizing circuit 30/50 is to be incremented by at least one discrete gain step. In one embodiment, the discrete gain step is approximately equal to 1.25. An M/N
circuit 96 receives peak history information from storage register 86 and determines if M peak values of N cardiac events are above the selected high threshold (7E hex). M/N circuit 96 provides a decrement signal on a line 98 to gain control cloclc circuit 94. M/N circuit 96 activates the decrement signal onl 5 line 98 when M out of N peak values are above the selected high threshold (7E
hex) to indicate that the gain of AGC/filter and digitizing circuit 30/50 is to be decremented by at least one discrete gain step. The decrementing discrete gain step is preferably equal to the incrementing discrete gain step and is approximately equal to 1.25 in one embodiment.
Gain control clock circuit 94 provides a gain control signal on a line 100 which controls the gain of AGC/filter and digitizing circuit 30/50 by causing the gain to be incremented or decremented in discrete gain steps based on the increment signal on line 92 and the decrement signal on line 98. The gainof AGC/filter and digitizing circuit 30/50 can be increased or decreased by a fixed number of steps or amount, or the level of the discrete gain step is optionally made programmable via microprocessor and memory 38. In addition, gain control and clock circuit 94 optionally causes increments or decrements of gain in multiple discrete gain steps. Since the increment signal on line 92 and the decrement signal on line 98 are never activated at the same time due to the dual low threshold (52 hex) and high threshold (7E hex), no arbitration circuitry is necessary to arbitrate between the increment or decrement signals to indicate A~N~ SHEET

SHEET

CA 02228~79 1998-02-04 which direction to proceed. Gain control circuit 32/52 preferably keeps the peakvalues of atrial or ventricle sensed cardiac events in approximately the upper one third of the dynamic range of A/D converter 62. As a result, the lower approximately two thirds of the dynamic range of A/D converter 62 is available 5 for sensing low amplitude signals such as occurring during fibrillation.
The above referenced number of M peak values is preferably odd to prevent lock-up of the AGC loop. For example, in a preferred embodiment of gain control circuit 32/52, M is equal to three and N is equal to four. In this embodiment, storage register 82 stores peak history information for four cardiac10 events in four corresponding storage locations each representative of whether the corresponding one of the last four values for peak values was below the selectedlow threshold (52 hex). In this embodiment, storage register 86 stores peak history information for four cardiac events in four corresponding storage locations each representative of whether the corresponding one of the last four 15 values for peak values was above the selected high threshold (7E hex).
The peak values in storage register 82 and storage register 86 are preferably updated at the beginning of a new refractory period for a previous sensed event. As new peak value information is acquired from comparator 80 the old peak history information is shifted one value to the right. If storage 20 registers 82 and 86 only contain four storage locations, the peak history values older than the last four cardiac events are shifted out of the registers to the right and lost.
In a preferred embodiment, M/N circuit 90 activates the increment signal on line 92 only if the stored peak value of the last cardiac event 25 and M - 1 peak values of the last N - 1 cardiac events previous to the last cardiac event are below the selected low threshold (52 hex). In this preferred embodiment, M/N circuit 96 activates the decrement signal on line 98 only if thestored peak value of the last cardiac event and M - 1 peak values of the last N - 1 cardiac events previous to the last cardiac event are above the selected high 30 threshold (7E hex).

'-~''''r'~D SHEET

CA 02228~79 1998-02-04 Gain control circuit 32/52 operates as described above to minimi7P the possibility of improper sensing by not allowing AGC/filter and digitizing circuit 30/50 to go to low sensitivity if large R-waves or P-waves are present or to go to full sensitivity in the presence of slow R-waves or P-waves.5 Improper sensing can cause therapy to be delivered to a patient at inappropriate times as a result of false indications of arrhythmia conditions. Oversensing is reduced because the full sensitivity of AGC/filter and digitizing circuit 30/50 is not reached between slow beats as a result of gain control circuit 32/52 keepingthe arnplified depolarization electrical activity in the upper approximately one10 third of the dynarnic range of A/D converter 62. The reduced oversensing greatly increases the comfort level of a patient having the cardioverter/defibrillator according to the present invention implanted in his or her body. Undersensing is reduced because minimum sensitivity will not occur due to a single large R-wave or P-wave.
In addition, as indicated above, gain control circuit 32/52 elimin~te~ the need for a high precision A/D converter implementation of A/D
converter 62, because the entire dynamic range of the incoming cardiac signal does not need to be spanned. Thus, in the preferred embodiment, A/D converter 62 is implemented in 8 bits or less. The dynamic range of the incoming cardiac signal from the atrial and/or ventricular channels of the heart ranges from 0. lmV to 25 mV representing a 250 to 1 dynamic range. Lower precision A/D
converters consume less power, convert the incoming analog signal to representative digital data more quickly, and allow more cost effective silicon processes to be l1tili7e~1 Moreover, m~nuf~cturability of the cardioverter/defibrillator is improved since no external parts are required to control the gain of the AGC/filter and digitizing circuit 30/50. Testing and characterization of the cardioverter/defibrillator is also improved since the digital logic of the gain control circuit 32/52 is easily fault graded.
AGC Turn Down Mechanism for Far Field Sensing The preferred embodiment of gain control circuit 32/52 illustrated in Figure 5 includes a far field sense circuit 102. Far field sense circuit 102 S~
~ ,c~Oc~-~

provides a solution to a possible AGC loop lock-up due to far field s~n~ing For example, when sensing events in the ventricle channel of the heart, P-waves, rt~ ,se~ g far field events, can be sensed during norrnal sinus rhythms at x; ~ ~ ~- ~ n ~ s~llsilivil~y. Under this example condition, instead of AGCing on the S R-wave peaks, which are clipped, the P-wave peak level ~It.qmAting with the clipped R-wave peak level combine to inhibit gain changes. The clipped R-wave peaks indicate that the R-wave peaks are above the m~X;~ digital value for a peak signal. In this case, the M of N algol;lhlll is never met in M/N circuit 96, which causes a lock-up condition in the AGC loop. Far field sense circuit 102 10 provides an additional gain decrease option to gain control circuit 32/52 in addition to the normal modes of operation to prevent this lock-up condition from occllrrin~
In the embodiment illustrated in Figure 5, two additional history storage locatioms are provided in storage register 86 to extend the peak history15 information to N + 2 storage locations. Far field sense circuit 102 responds to the last N + 2 sensed events stored in storage register 86 to d~qterrnine if thestorage information sllt~ tes between clipped peaks and non-clipped peaks for the last N + 2 sensed events. Far field sense circuit 102 detPrmin~s that a peak is clipped when the peak is at the m~x; ~ " . value (7F hex) which c~lle~l,ollds to20 peak values greater than the high threshold value used by colll~ alol 80 (7E
hex). Far field sense circuit 102 provides a decrement signal on a line 104 to gain control clock circuit 94. Far field sense circuit 102 activates the decrement signal on line 104 when the peak history infor nation in storage register 86 ;s b~lw~ell clipped peaks and non-clipped peaks for the last N + 2 sensed 25 events. In one embodiment, the decrement signal on line 104 indicates that gain of AGC/filter and digitizing circuit 30/50 is to be decr~mented by one discrete gain step, but can ~ltern:~tively in~lic~t~ any number of discrete gain step clu~ngrc, Far field sense circuit 102 operates in the cases where the actual 30 depolarization of the incoming cardiac signal is clipped to prevent the digital AGC loop from locking up under the condition of far field sensed events. If far -- CA 02228~79 1998-02-04 field events are detected in the above manner, the gain of AGC/filter and digitizing circuit 30/50 is decreased to the point that far field events no longer are sensed. Previous cardioverter/defibrillator devices all oversense (double count) under this far field sensing condition. The far field sense circuit 102 5 greatly improves sensing discrimination by minimi7:ing or substantially elimin~ling oversensing in the presence of far field events. Accordingly, the cardioverter/defibrillator according to the present invention provides a patientand his or her physician a cardioverter/defibrillator which senses the R-wave depolarizations more reliably.
10 Slow Gain Jump Back for AGC
Previous gain circuitry reaches maximum sensitivity in a single cardiac cycle. Unlike previous gain ~;hcuilly, the slow gain circuitry accordingto the present invention described above makes discrete step gain changes of oneor more discrete gain step per cardiac depolarization cycle, so that full sensitivity 15 of the AGC/filter and digitizing circuit 30/50 is not reached between cardiacdepolarizations, which can cause undersensing of cardiac events. As illustrated in Figure 5, additional circuitry is preferably added to gain control circuit 32/52 to prevent undersensing of cardiac events.
Exception ~;hcuill y 106 detects any one of three conditions which 20 indicate that the gain of the AGC/filter and digitizing circuit 30/50 is to be set to a selected relatively high sensitivity. Exception circuitry 106 provides a set gain signal on a line 108 to cause the gain of AGC/filter and digitizing circuit 30/50 to be set to the selected relatively high sensitivity when any of the three conditions occur. The first condition occurs when a cardiac event is not detected 25 for a selected time period (i.e., a R-wave or P-wave depolarization is not sensed for the selected time period). Typically, the selected time period is equal to approximately 1.5 seconds,.corresponding to a heart rate of less than 40 beats per minute. The second condition occurs after the cardioverter/defibrillator delivers a shock pulse. The third condition occurs after the cardioverter/defibrillator 30 delivers a pacing pulse.

~ S~

CA 02228~79 1998-02-04 In any of the three conditions, it is desirable to prevent undersensing by setting the gain of the AGC/filter and digitizing circuit 30/50 to the selected relatively high sensitivity to quickly increase the sensitivity of the cardioverter/defibrillator. A/D converter 62 (shown in Figure 2) typically S operates in bands of an approximately 10:1 dynamic range. The combined 10:1 dynarnic range bands create a total 250:1 dynamic range of A/D converter 62.
The three exception conditions are conditions where A/D converter 62 needs to operate near maximum sensitivity, or in other words, near the upper portion of the highest 10:1 dynamic range band to adequately prevent undersensing.
In the preferred embodiment, the selected relatively high sensitivity is two gain steps from a maximum sensitivity to prevent mistaking P-wave depolarizations and T-wave repolarizations for R-wave depolarizations. If the selected relatively high sensitivity creates a clipped signal on the following depolarization, having its peak at the maximum value (7F hex), as indicated fromcomparator 80 on line 88, a jump back compare circuit 110 activates a line 114 to a two input OR gate 1 16 to indicate that the gain is to be reduced by an offset value stored in offset register 112. OR gate 116 provides a decrement signal on an enable line 120 to gain control clock circuit 94 which is activated when either of the two inputs to the OR gate are activated to indicate that the gain of AGC/filter and digitizing circuit 30/50 is to be decremented by at least one discrete gain step during the current refractory period. The offset value stored in register 112 is preferably programmable and is provided to gain control clock circuit 94. In one embodiment, the offset value is programmed to equal three discrete gain steps.
If the peak value of the digital cardiac data on line 72 is still clipped on the next depolarization after the gain has been decreased by the offset value stored in offset register 112, comparator 80 indicates on line 88 that thepeak of the cardiac signal is still clipped. Jump back compare circuit 110 then indicates that the gain is to be decremented by at least one discrete gain step by activating a line 118 to the other input of OR gate 116, which correspondingly activates enable line 120 to gain control clock circuit 94. If the peak value is still ~E~D~ S~

clipped, normal AGC action as described above resllmPs This two staged back offm~ch~ni~m after a jump out or escape to the selected relatively high sensitivity due to lack of sensing reduces ov~ resl-lting from the clipped peak of the cardiac signal.
S If the peak value of the digital cardiac data on line 72 is not clipped on the first depol~ri7~tion after the gain is set to the relatively highsensitivity, but the peak value is clipped on the second depolarization after the gain is set to the relatively high sensitivity by having its peak at the m~ -lu"l value (7F hex), as indicated from cu~ aldLI l 80 on line 88, jump back compare circuit 1 10 activates line 114 to two input OR gate I l 6 to indicate that the gain is to be reduced by the offset value stored in offset register 11 2. OR gate 116 provides the decrement signal on enable line 120 to gain control clock circuit 94 which is activa~ed when either of the two inputs to the OR gate are activated toinllis~te that the gain of AGC/filter and digitizing circuit 30/50 is to be decremented by at least one discrete gain step during the current refractory period. If the peak value is still clipped, normal AGC action as described aboveresl-m~os This situation, where the peak value of the first detect~cl depolarization is not clipped and the peak value of the second detect~d depolarization is clipped after the gain is set to the relatively high sensitivit.v, results when the first depolarization represents a far field sensed event such as described above. For example, when sensing events in the ventricle channel of the heart, P-waves, f s~ g far field events, can be sensed during normal sinus rhythms at m~ sensitivity.
Threshold Ten~latin~ for a Fast Di~ital AGC Circuit Figure 6 illu~l~dles, in timing diagram form, the variable sensing threshold g~;n~.dl~d by template generation circuit 36/56 and provided to detection circuil~ 34/54. The variable sensing threshold is indicated by line 130.
As illustrated, the variable sensing threshold 130 follows a piecewise linear - approximation of an exponential decay curve with minim~l error between steps.
The template generation circuit 36/56 forces the variable sensing threshold 130 to rapidly follow the maximum peak level of the ~1igi~i7~d cardiac data. When CA 02228~79 1998-02-04 the incoming digitized cardiac data is greater than the current sensing threshold, template generation circuit 36/56 raises the variable sensing threshold 130 to apeak threshold value approximately equal to the peak value of the incoming digitized cardiac data as indicated at time T(0). After a shock or pace pulse is5 delivered by the cardioverter/defibrillator, template generation circuit 36/56 sets the variable sensing threshold 130 to a selected relatively high threshold value.
The selected relatively high threshold value is preferably 7E hex in the exarnple embodiment or one binary number below the maximum value of the variable threshold.
The variable sensing threshold 130 remains at the peak threshold value through a refractory or a portion of a refractory period indicated at l 32.
When the cardioverter/defibrillator is operating in pacing mode, the period indicated at 132 is a programmed paced refractory period that is selected by thephysician and prograrnmed into the cardioverter/defibrillator, such as the 15 paced/shock refractory period indicated at 81 is Figure 4. When the cardioverter/defibrillator is operating in shocking mode, the period indicated at 132 is a shock refractory period, such as the paced refractory period indicated at 81 is Figure 4. When the cardioverter/defibrillator is operating in sensing mode, the period indicated at 132 is a portion of a sensed refractory period, such as the 20 absolute refractory period indicated at 78 is Figure 4. In addition to the refractory or the portion of a refractory period indicated at 132, the variable sensing threshold does not begin to decay from the peak threshold value attainedat tirne T(0) for an additional drop time indicated at 134. The drop time is a normal template hold time for the peak converter circuitry of template generation 25 circuit 36/56, and is empirically determined. A suitable value for the drop time in one embodiment is approximately 13.7 msec.
After the refractory period and the drop time have elapsed, at time T(1,0) the variable sensing threshold 130 drops by an initial drop percentage, indicated by arrows 136. The initial drop percentage is preferably approximately30 25% of the peak threshold value so that the level of the variable sensing threshold obtained at time T(1,0) is approximately 75% of the initial peak ~ 0 CA 02228~79 1998-02-04 threshold value. As indicated at time T(1,1), the variable sensing threshold starts to decay in discrete steps such as indicated at 138. The step time size is representatively indicated by arrows 140 between time T(1,1) and T(1,2). The level of the variable sensing threshold 130 decays from a percentage of the peakS threshold value to step over wide depolarizations or T-waves in the incoming electrical activity.
In the preferred embodiment, template generation circuit 36/56 drops the variable sensing threshold 130 in step groups comprising multiple discrete steps. In the embodiment illustrated in Figure 6, the step group size is 10 four. Each step group decreases the variable sensing threshold by a defined percentage, as indicated by arrows 142 for a four step group between time T(1,0)and T(2,0), arrows 144 for a four step group between T(2,0) and T(3,0), and arrows 146 for a four step group between T(3,0) and T(4,0). The defined percentage for each step group is preferably approximately 50%. For example, 15 in the preferred embodiment, the value of the variable sensing threshold at time T(2,0) is approximately 50% of the value of the variable sensing threshold at time T(1,0), and the value of the variable sensing threshold at time T(3,0) is approximately 50% of the value of the variable sensing threshold at time T(2,0) or 25% of the value of the variable sensing threshold at time T(1,0).
When the variable sensing threshold 130 decays to a programmable final value, as indicated at 148, template generation circuit 36/56holds the variable sensing threshold at the programmable final value until a newsensed event occurs. The programmable final value is programmable to compensate for noise which is inherent in the sense amplifiers and other AGC
25 system circuits of the AGC loop.
The initial drop percentage to achieve approximately 75% of the peak threshold value, and the four discrete steps in each step group to drop thevariable sensing threshold to approximately 50% of the level of the start of thefour-step group realizes a piecewise geometric progression linear approximation 30 representing an exponential decay curve with minim~l error between piecewise CA 02228~79 1998-02-04 steps. Since the sensing threshold drops in discrete steps as indicated at 138, integer math can be utilized in template generation circuit 36/56. For example in the embodiment of template generation circuit 36/56 illustrated in Figure 6, floating point nurnbers are not required because the maximum difference/error S between any two discrete steps in a four step group is one bit. The other embodiments can be extended to use any size integer value or nurnber of steps orstep groups to achieve the linear approximation of the exponential decay curve.
In fact, floating point numbers are optionally used, but are not desirable because of the increased silicon area needed to implement floating point logic circuits. In 10 addition, by implementing the template generation circuit witn integer values, the resulting template generation circuit consumes a relatively small amount of power.
A preferred algorithm for calculating the drop in amplitude for each of the discrete steps is shown in TABLE I below.

DE~) S~

W O 97/06852 PCT~US96/13205 T~BT,~, I

T~TFRVAL STFP C~T ,CUT ~TION

TEMP = T(0) - T(0) /2 + T(0) /4 TEMPl = T(X-1,0) /2 T(X,0) = IF (X = 1) THEN
IF (FINAL THRESHOLD > TEMP) THEN
FINAL ELSE TEMP
ELSE
IF (FINAL THRESHOLD > TEMPI) THEN
FINAL ELSE TEMPI

TEMP = T(X,0) - T(X,0) /4 + T(X,0) /8 T(X,l) = IF (FINAL THRESHOLD > TEMP) THEN FINAL
ELSE TEMP

TEMP = T(X,0) - T(X,0) /2 + T(X,0) /4 T(X,2) = IF (FINAL THRESHOLD > TEMP) THEN FINAL
ELSE TEMP

TEMP = T(X,0) - T(X,0) /2 + T(X,0) /8 T(X,3) = IF (FINAL THRESHOLD > TEMP) THEN FINAL
ELSE TEMP

Where:
T0 = PEAK THRESHOLD VALUE
- T(X,1.. 3) = One of Four Steps X = 1,2,3,4 - Decay Period CA 02228~79 1998-02-04 W O 97/06852 PCT~US96/13205 Referring to TABLE I above, in the interval T(X,0), TEMP is calculated to 75% of the initial peak threshold value, and TEMPl is calculated to 50% of a previous step group value. If the step group is the first drop from thepeak threshold value, then T(l ,0) is equal to TEMP or 75% of the peak value. In5 crlccP~ive drops, T(X,0) is equal to TEMPI or a 50% drop from the level at the l~cgi ~ g of the previous step group.
In all of the T(X, 1), T(X,2), and T(X,3) intervals, the variable sensing threshold obtains the TEMP value unless the TEMP value is less than FrNAL TEIRESHOLD which is the final pro~d~ able value indicated at 148 in 10 Figure 6. For exarnple, in the T(X,I) interval, T(X,1) is set to TEMP which is calculated to 87.5% of the T(X,0) value. In the T(X,2) interval, T(X,2)is set toTEMP which is calculated to 75% of the T(X,0) value. In the T(X,3) interval, T(X,3) is set to TEMP which is calculated to 62.5% of the T(X,0) value.
A logical block tii~gr~m of a pl~efe.,~d embodiment of template 15 generation circuit 36/56, which uses integer values for calculating the variable sensing threshold, is illustrated in Figure 7. A peak detection circuit 160 detects the peak value of the digitized cardiac data provided on line 72 from AGC/filterand ~ligiti7.ing circuit 30/50. Peak detection circuit 160 provides a peak threshold value which is equal to the peak value of the digitized data to a threshold register 162 if the ~ iti7~d peak is greater than the current threshold value. Threshold register 162 stores and provides the current variable sensing threshold on line 164 to the detection circuit 34/54. Peak detection circuit 160 also provides thepeak threshold value to T(X,0) register 166.
If the step group is not the first drop from the peak threshold 25 value the TEMP 1 calculation must be implernl?nted for the T(X,0) interval of the discrete step calculation algorithm in TABLE I above. To implement the TEMP1 calculation, the T(X-1,0) value stored in the T(X,0) register 166 from the previous step group is divided by 2 through a hard shift of one to the right as indicated by line 168 to place the shifted data in both the threshold register 162 30 and the T(X,0) register 166.

W O 97/06852 PCT~US96/13205 T(X,0) register 166 provides its ~ lly stored value to a subtraction circuit 170 and a shifter 172. ShiflLer 172 provides either a divide by 2 or a divide by 4 calculation by shifting the current T(X,0) value by one bit or two bits to the right, respectively. Subtractor 170 subtracts the value stored in ~ 5 the T(X,0) register 166 from a shifted output provided from shifter 172. The shifted output of shifter 162 is also provided to a shifter 174. Shifter 174 provides an ~ 1ition~1 divide by 2 or divide by 4 through shifts of 1 bit or 2 bits to the right, re~e~ ely. A difference output of subtractor 170 is provided to anadder 176. A shifted output of shifter 174 is provided to the other input of adder 10 176. Adder 176 adds the difference output of subtractor 170 and the shifted output of shifter 174 and provides the added value to threshold register 162.
The shifters 172 and 174 can, in combination, achieve shifts of 1, 2,3, or 4 bits to produce divide by 2, divide by 4, divide by 8, or divide by 16calculations. The TEMP calculations required for the T(X,0), T(X,l), T(X,2), 15 and T(X,3) intervals of the discrete step calcul~tion algo~ in TABLE I above are all achieved through shifters 172 and 174 in combination with subtractor 170and adder 176. Shifters 172 and 174 Ç~ te the desired divide by values which are then l,r~ .ly comhin~l according to the algo,it~l.n in TABLE I with subtractor 170 and adder 176.
A final threshold register 178 stores the programmable final value, in-1ic~1ecl at 148 in Figure 6, ofthe variable sensing threshold. The progl~l~ able final value is provided to a threshold col,lp~dtor 180. Threshold COlll~dlOl 180 col,.pales the programmable final value stored in final thresholdregister 178 with the current variable sensing threshold value on line 164.
Threshold Culll~dtOl 180 indicates to threshold register 162, on a line 182, whether the current variable sensing threshold value is greater than the progr~mm~hle final value. If the program~ mable final value is greater than the calculated sensing threshold value, then the final value is stored in threshold register 162. The sensing threshold value stays at the final value until the h~co~ lgrli~;iti7~ cardiac data exceeds the final value in(licatin~ a new sensedevent. In fact, a new sensed event occurs any time the incoming ~ iti7P~Cl CA 02228~79 1998-02-04.

cardiac data peak value exceeds the current variable sensing threshold value on line 164. With the new sensed event, the variable sensing threshold obtains a new T(0) peak threshold value equal to the peak value of the sensed depolarization in the digitized cardiac data.
S The above described threshold templating algorithm for a fast digital AGC system is completely contained in digital logic as implemented in the preferred embodiment. The digital logic implementation is easily characterized, tested, and achieves repeatable results. In addition, external parts are elimin~tecl from the silicon chip implementation of the AGC circuitry to 10 reduce cost and increase the m~nllf~cturability of the AGC silicon chip. Testing and characterization of the cardioverter/defibrillator devices is uniform from one device to another. In this way, it is easier for the physician to determine how to implement the cardioverter/defibrillator device in a patient, because the devicereacts consistently from one unit to another.
15 Tailorable AGC Decav Rate No single decay rate (attack rate) is optimal for all operating conditions of a cardioverter/defibrillator with pacing capability for the above described fast response AGC circuit. The typical operating conditions encountered include bradycardia pacing, tachyrhythmia sensing, and normal 20 sinus sensing. Therefore, the step time size indicated by arrows 140 in Figure 6 is programmable to achieve a tailorable AGC decay rate for the variable sensing threshold 130. In this way, by varying the step time size 140 for each of the defibrillator's operating conditions, the decay rate is customized to optimally meet the selected operating condition.
For normal sinus sensing, a single attack rate is utilized that covers most of the incoming cardiac signals. In one embodiment, the step time size 140 is set to 29.3 mSec/step to achieve the normal sinus sensing decay rate.
Tachyrhythmia sensing is a special condition under which a fast response rate is desirable in order to properly track the higher tachyrhythmia 30 rates, such as during fibrillation or tachycardia. This is especially true in the A~E~E~ SHE~T

CA 02228~79 1998-02-04 atrium of the heart, where tachyrhythmia rates run in excess of 300 beats per minute. In one embodiment, step size 140 is set to approximately 17.5 mSec/step for atrial tachyrhythmia conditions, and is set to approximately 23.5 mSec/step for ventricle tachyrhythmia conditions. By switching to this faster 5 decay rate for tachyrhythmia conditions, cases of undersensing a tachyrhythmia condition which needs to be treated is reduced.
Bradycardia pacing is a special operating condition wherein the decay rate of the sensing threshold is tied to the bradycardia pacing rate to help minimi7e oversensing and undersensing conditions. In prior 10 cardioverter/defibrillator devices with pacing capability, the sensing template attack rate is fixed. Under situations of high pacing rates, the cardioverter/defibrillator with pacing capability ~ltili7ing AGC according to the present invention does not have time to decay to maximu}n sensitivity. If the decay rate is not sufficiently sped up along with the high pacing rates 15 undersensing occurs and the cardioverter/defibrillator continues pacing in the presence of fibrillation. With the decay rate varied as a function of the bradycardia pacing rate under bradycardia pacing conditions, the decay rate is sufficiently sped up to enable the cardioverter/defibrillator according to the present invention to sense and properly respond to the fibrillation condition. In 20 addition, when pacing rates are low, a longer decay rate is desirable to minimi7e the possibility of oversensing.
The formula for calculating the post pace template step time size 140 for bradycardia pacing conditions is as follows:

CA 02228~79 1998-02-04 32 ..
STEP TIME SIZE =
(CYCLE LENGTH - REFRACTORY - DROP TIME - MINIMUM TIME) /X
where:
CYCLE LENGTH = pacing cycle length REFRACTORY = prograrnmed paced refractory DROP TIME = normal template hold time for peak converter (approximately 13.7 mSec in a preferred embodiment) MINIMUM TIME = minimum time allowed for template at final value (approximately 100 mSec in a preferred embodiment) X = number of steps to go from seed value to final value (equal to 12 steps in the embodiment illustrated in Figure 6) Referring to Figure 6, the cycle length is equal to the pacing cycle length or from time T(0) to T(0) between each pacing pulse. The paced refractory period is indicated by arrows 132. The drop time is indicated by arrows 134. The time the variable sensing threshold is at the programmable final20 value before the next pacing pulse is indicated by arrows 150. Since multiplepacing rates are assigned the same step size, the time indicated at 150 varies from approximately 100 mSec to 200 mSec in the embodiment illustrated. The minimum time is the minimum time allowed for the time indicated by arrows 150, or approximately 100 mSec. X represents the 12 steps (i.e., the 3 X four 25 step groups) to go from the peak sensing value at time T(0) to the programmed final value of the variable sensing threshold achieved at T(4,0).
A look-up table stored in microprocessor and memory 38 is formed by dividing the cycle length by 64, which results in a shift of six bits to the right. In one implementation, the cycle length is equal to 12 30 bits, which results in six bits being shifted off in the divide by 64 formation of the look-up table in microprocessor and memory 38, resulting in 64 CA 02228~79 1998-02-04 entries in the look-up table. Thus, the current cycle length is divided by 64 toindex the look-up table to access the values stored in the look-up table corresponding to the above step time size forrnula.
The digital embodiment of the AGC loop as described above S allows the above described firrnware implemented in the look-up table in the microprocessor and memory 38 to dynamically adjust the sensing characteristics of the cardioverter/defibrillator according to the present invention. By sensinghigh rates differently than low rates, the tailorable AGC decay rate can be utilized to orthogonally optimize sensing characteristics of bradycardia and 10 tachyrhythmia signals, which have mutually exclusive sensing requirements. Inthis way, the physician controls a better-behaved cardioverter/defibrillator. Inaddition, patient comfort is increased, due to reducing oversensing and undersensing of treatable arrhythrnia conditions in the patient.
Interaction of Digital AGC Using Separate Gain Control and Threshold 15 Templating Figure 8 illustrates in tirning diagrarn form depolarization cycles in the electrical activity of the heart. The incoming electrical activity at input/output terrninals 22 or 42 is indicated by waveform 200. The filtered and gain controlled digitized cardiac signal is indicated by waveform 202. The 20 variable sensing threshold is indicated by waveform 204. The absolute value of the digitized and gain controlled cardiac signal is indicated by waveform 206 underneath the variable sensing threshold waveform 204. The refractory period is indicated by waveform 208. The discrete stepped slow gain is indicated by waveform 210.
As indicated by waveform 204, the variable sensing threshold waveform responds to the absolute value of the t1i~iti7ed cardiac signal to assurne the peak value of the digitized cardiac signal. The variable sensing threshold then decays according to a piecewise linear approximation of an exponential decay curve to step over wide depolarizations or T-waves.

~ lENDED SHEET

~ CA 02228~79 1998-02-04 , , The influence of the slow gain control on the fast templating circuit is illustrated at time 212. As is indicated, the gain is decreased at time 212, which correspondingly results in a reduced filtered and gain controlled digitized cardiac signal indicated at 202, which correspondingly reduces the 5 variable sensing threshold indicated at 204 as the variable sensing threshold follows the peak value of the absolute value of the digitized and gain controlled cardiac signal indicated at 206.
Conclusion By ntili7in~ two independent loops in a cardioverter defibrillator l O with pacing capability which are both implementing digital logic circuits, the AGC response is effectively moved from analog circuits into the digital logic circuits, where it is easier to test and characterize. Design of the sense amplifier is simplified, due to the digital control of the sense amplifier. It is easier to test and characterize the analog sense amplifier, since the AGC circuitry is no longer 15 in the analog domain. The cardioverter/defibrillator device is more uniform from device to device, which greatly increases the physician's ease of predicting device behavior. In addition, the patient comfort is increased due to reduced oversensing and undersensing.
Although specific embodirnents have been illustrated and 20 described herein for purposes of description of the preferred embodiment, it will be appreciated by those of ordinary skill in the art that a wide variety of alternate and/or equivalent implementations calculated to achieve the same purposes may be substituted for the specific embodiments shown and described. Those with skill in the mechanical, electro-mechanical, electrical, and computer arts will 25 readily appreciate that the present invention may be implemented in a very wide variety of embodiments. This application is intended to cover any adaptations orvariations of the preferred embo~1iment.c discussed herein. Therefore, it is manifestly inten~le~l that this invention be limited only by the claims and the equivalents thereof.

AMENDED SiiEE~

Claims (26)

WHAT WE CLAIM IS:

1. A method of automatically controlling a gain of a cardioverter/defibrillator which receives electrical activity of the heart and provides shock pulses in response thereto, the electrical activity is amplified according to a variable gain and cardiac events are detected representing depolarizations in the amplified electrical activity which exceed a sensitivity threshold, the method being characterized by the steps of:
storing history information representative of the amplified electrical activity of a first selected number (N) of cardiac events;
increasing the variable gain if a second selected number (M) of peak values of the amplified electrical activity of the N cardiac events are below a selected low threshold, wherein M is an odd integer number equal to or greater than 3 and N is an integer number equal to or greater than M; and decreasing the variable gain if M peak values of the amplified electrical activity of the N cardiac events are above a selected high threshold.

2. The method of claim 1 wherein the variable gain is increased in the increasing step if a stored peak value of the last cardiac event and M-1 peak values of the last N-1 cardiac events previous to the last cardiac event are below the selected low threshold and the variable gain is decreased in the decreasing step if the stored peak value of the last cardiac event and M-1 peak values of the last N-1 cardiac events previous to the last cardiac event are above the selected high threshold.

3. The method of claim 1 further comprising the step of setting the variable gain to a selected relatively high sensitivity, which is at least one discrete gain step from a maximum sensitivity, based on a certain condition occurring which substantially invalidates the history information.

4. The method of claim 1 further comprising the step of decreasing the variable gain if the peak values of the amplified electrical activity of a thirdselected number of cardiac events alternate between clipped peak values and non-clipped peak values.

5. A system for automatically controlling a gain of a cardioverter/defibrillator, which receives electrical activity of the heart and provides shock pulses in response thereto, the system including an amplifier (30/50) for amplifying the electrical activity of the heart according to a variable gain, and a cardiac depolarization detector (34/54) for detecting depolarizations in the amplified electrical activity of the heart and providing a detect signal representing a cardiac event indicative of a depolarization when the amplified electrical activity exceeds a sensitivity threshold, the system characterized by:
storage means (82/86) for storing history information representative of values of the amplified electrical activity of a first selected number (N) of cardiac events; and gain controller (32/52) for adjusting the variable gain based on the stored history information by increasing the variable gain if a second selected number (M) of peak values of the N cardiac events are below a selected low threshold and by decreasing the variable gain if M peak values of the N cardiac events areabove a selected high threshold, wherein M is an odd integer number equal to or greater than 3 and N is an integer number equal to or greater than M.

6. The system of claim 5 wherein the stored history information is peak history information.

7. The system of claim 5 wherein the gain controller adjusts the variable gain in discrete steps.

8. The system of claim 7 wherein the gain controller increases the variable gain if a stored peak value of the last cardiac event and M-1 peak values of the last N-1 cardiac events previous to the last cardiac event are below the selected low threshold and decreases the variable gain if the stored peak value of the last cardiac event and M-1 peak values of the last N-1 cardiac events previous to thelast cardiac event are above the selected high threshold.

9. The system of claim 7 wherein the storage means comprises a first group of storage locations which store peak history information indicating if the peakvalues are below the selected low threshold and a second group of storage locations which store peak history information indicating if the peak values areabove the selected high threshold.

10. The system of claim 5 wherein the peak history information from the previous cardiac event is updated at the beginning of a new refractory period caused by a cardiac event.

11. The system of claim 5 wherein the gain controller includes means responsive to the detect signal to set the variable gain to a selected relatively high sensitivity, which is at least one discrete gain step from a maximum sensitivity, based on a certain condition occurring which substantially invalidates the peak history information.

12. The system of claim 5 wherein the storage means is capable of storing peak history information representative of peak values of the amplified electrical activity of a second selected number of cardiac events, and the system further comprises:
gain turndown means responsive to the stored peak history information to decrease the variable gain by at least one discrete step if the peak values of the amplified electrical activity of the second selected number of cardiac events alternate between clipped peak values and non-clipped peak values.

13. The method of claim 4 further comprising the steps of:

digitizing the amplified electrical activity; and comparing the digitized amplified electrical activity to a maximum peak digital value and indicating that a peak value is clipped when the peak value ofthe digitized amplified electrical activity is equal to the maximum peak digitalvalue.

14. The system of claim 12 further comprising:
a analog to digital converter to digitize the amplified electrical activity;
and a comparator for comparing the digitized amplified electrical activity to a maximum peak digital value and indicating that a peak value is clipped when the peak value of the digitized amplified electrical activity is equal to the maximum peak digital value.

15. The method of claim 3 wherein the certain condition occurs when a cardiac event is not detected within a selected time period.

16. The method of claim 3 wherein the certain condition occurs after a shock pulse is delivered by the cardioverter/defibrillator.

17. The method of claim 3 wherein the method automatically controls a gain of a cardioverter/defibrillator having pacing capability, and wherein the certain condition occurs after a pacing pulse is delivered by the cardioverter/defibrillator.

18. The method of claim 3 further comprising the step of decrementing the variable gain from the selected relatively high sensitivity by a selected number of discrete gain steps if the setting of the variable gain to the selected relatively high sensitivity creates a clipped peak value of the amplified electrical activity on the following detected cardiac event.

19. The method of claim 18 further comprising the step of further decrementing the variable gain by at least one discrete gain step if the peak value of the amplified electrical activity is still clipped on the second detected cardiac event following the setting of the variable gain to the selected relatively highsensitivity.

20. The method of claim 3 further comprising the step of decrementing the variable gain from the selected relatively high sensitivity by a selected number of discrete gain steps if the setting of the variable gain to the selected relatively high sensitivity does not create a clipped peak value of the amplified electrical activity on the following detected cardiac event and does create a clipped peak value of the amplified electrical activity on the second detected cardiac event following the setting of the variable gain to the selected relatively high sensitivity.

21. The system of claim 11 wherein the certain condition occurs when a cardiac event is not detected within a selected time period.

22. The system of claim 11 wherein the certain condition occurs after a shock pulse is delivered by the cardioverter/defibrillator.

23. The system of claim 11 wherein the system automatically controls a gain of a cardioverter/defibrillator having pacing capability, and wherein the certain condition occurs after a pacing pulse is delivered by the cardioverter/defibrillator.

24. The system of claim 11 further comprising means for decrementing the variable gain from the selected relatively high sensitivity by a selected number of discrete gain steps if the setting of the variable gain to the selected relatively high sensitivity creates a clipped peak value of the amplified electrical activity on the following detected cardiac event.

25. The system of claim 24 further comprising means for decrementing the variable gain by at least one discrete gain step if the peak value of the amplified electrical activity is still clipped on the second detected cardiac event following the setting of the variable gain to the selected relatively high sensitivity.

26. The system of claim 11 further comprising means for decrementing the variable gain from the selected relatively high sensitivity by a selected number of discrete gain steps if the setting of the variable gain to the selected relatively high sensitivity does not create a clipped peak value of the amplified electrical activity on the following detected cardiac event and does create a clipped peak value of the amplified electrical activity on the second detected cardiac event following the setting of the variable gain to the selected relatively high sensitivity.