Implantable medical device with self-correlation means and with analysis means for estimating cardiac rate
Abstract
Autocorrelation enhancement and implementation are described. In particular, certain examples demonstrate the use of peak selectors to identify peaks of autocorrelation functions that serve as candidate heart rates for implantable medical devices. The approach can allow an alternative calculation of heart rate in an implantable medical device, either as a stand-alone rate detector or as a double check for other rate calculations.

Term
8.9 yearsto projected expiry
Projected expiry 13 August 2035, counted from filing; an application has no term until it is granted.
- Priority
- Filed
- Published
- Today
- Projected expiry
15 claims: 4 independent, 11 dependent
- 1心臓信号の反復的な分析のために構成されている植え込み型の医療用デバイスシステム(12、14、32、34)であって、 心臓信号を感知するための複数の電極(16、18、20、36、38、40、42)と、 該感知された心臓信号から自己相関関数を発生させるための自己相関手段であって、該自己相関関数は、ラグ深さの関数として振幅を有する、自己相関手段と、 該自己相関関数の中の振幅ピークを識別するとともに心拍数の第1の推定値、および、少なくとも低い信頼度または高い信頼度の格付けを有する第1の関連の信頼度を見出すためのピークセレクタ手段と、を備え、 該ピークセレクタ手段は、所与の反復計算の前記自己相関関数における第1のラグ深さを有する選択されたピークに関して、少なくとも1つの追加的なピークが、該第1のラグ深さの倍数である第2のラグ深さに出現するかどうかを決定するためのピケットテスト手段を含んでおり、 該ピークセレクタ手段は、さらに、候補ピークを選択し、推定される心拍数を計算することに関する該候補ピークの適切性を決定するための候補選択手段を含んでおり、 該候補選択手段は、前記自己相関関数のピークを使用して多数の候補ピークを識別するように動作可能であり、 前記候補選択手段は、第1の候補ピークとして、以下、すなわち、 前記識別された候補ピークのうちの最小のラグ深さを有する候補ピーク、または、 最小のラグ深さを備える前記候補ピークのものよりも、少なくとも第1のマージンだけ大きい振幅を有する候補ピークであって、レート閾値を超過する心拍数に対応する、候補ピーク、のうちの1つを選択するように動作可能であり、 前記ピークセレクタ手段は、前記候補選択手段を使用し、1つまたは複数の候補ピークを識別するように構成されており、また、前記ピケットテスト手段を使用し、任意の候補ピークが心拍数を推定するのに適切であるかを決定し、そうである場合には、推定された心拍数を報告するように構成されている、植え込み型の医療用デバイスシステム(12、14、32、34)。
- 2前記ピークセレクタ手段は、前記候補ピークのいずれかが、すべての他のピークを少なくとも第2のマージンだけ超過しているかどうかを決定するためのドミナントピークテスティング手段をさらに含み、前記ピークセレクタ手段は、前記ドミナントピークテスティング手段を使用し、前記ピケットテスト手段によって適切であると見出される候補ピークがない場合には、心拍数を推定するのに適切なピークを識別しようと試みるように動作可能である、請求項1に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 3前記ドミナントピークテスティング手段が、少なくとも前記第2のマージンだけすべての他のピークを超過しているドミナントピークを識別し、前記ドミナントピークが、ドミナントピークレート閾値を下回る心拍数に対応する場合には、前記ピークセレクタ手段は、前記ドミナントピークに対応する前記心拍数を推定された心拍数として報告する、請求項2に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 4前記植え込み型の医療用デバイスシステム(12、14、32、34)は、 前記電極からの電気信号を閾値と比較することによって、心イベントを検出し、それによって、複数のR波検出および結果として得られるレート推定値を発生させるためのR波検出手段と、 該R波検出手段およびトラッキング手段のそれぞれからの結果をとり、治療が必要とされるかどうかを決定するための決定手段と、をさらに備え、 該決定手段は、ドミナントピークに基づいて前記ピークセレクタ手段によって報告される推定された心拍数を、該R波検出手段によって発生させられるレートよりも信頼性が高いものとして受け入れるように構成されている、請求項3に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 5前記候補選択手段は、前記自己相関関数の中の所定の量の最大ピークを選択するように構成されており、 前記ピークセレクタ手段は、頻拍レートに対応する深さにおける前記自己相関関数の中のピークが、前記自己相関関数の中の前記最大ピークの第3のマージンの中にあるかどうかを決定するための大きい頻拍ピークチェック手段をさらに含み、そうである場合には、前記ピークセレクタ手段は、前記大きい頻拍ピークチェック手段によって識別された前記ピークを前記ピケットテスト手段に提出し、それが心拍数を推定するのに適切であるかどうかを決定するように構成されている、請求項1または2に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 6心拍数推定値を発生させるべく前記ピークセレクタ手段の出力をトラッキングするためのトラッキング手段をさらに備える、請求項1乃至3のいずれか1項に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 7報告閾値よりも大きい前記自己相関関数の中の任意のピークを識別し、前記トラッキング手段に報告するための報告手段をさらに備える、請求項6に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 8前記自己相関関数の中の最大ピークを識別し、該最大ピークの閾値パーセンテージよりも大きい前記自己相関関数の中の任意のピークを前記トラッキング手段に報告するための報告手段をさらに備える、請求項6に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 9前記電極からの電気信号を閾値と比較することによって、心イベントを検出し、それによって、複数のR波検出および結果として得られるレート推定値を発生させるためのR波検出手段と、 該R波検出手段および前記トラッキング手段のそれぞれからの結果をとり、治療が必要とされるかどうかを決定するための決定手段と、をさらに備える、請求項6乃至8のいずれか1項に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 10前記電極からの電気信号を閾値と比較することによって、心イベントを検出し、それによって、複数のR波検出および結果として得られるレート推定値を発生させるためのR波検出手段と、 該R波検出手段および前記ピークセレクタ手段のそれぞれからの結果をとり、治療が必要とされるかどうかを決定するための決定手段と、をさらに備える、請求項1乃至3のいずれか1項に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 11前記決定手段は、前記ピークセレクタ手段によって報告されるレート推定値が、心拍数を推定するのに適切であるとして前記ピケットテスト手段によって識別された候補ピークに基づいている場合には、前記ピークセレクタ手段からの結果を、前記R波検出手段からの結果よりも信頼性が高いものとして扱うように構成されている、請求項10に記載の植え込み型の医療用デバイスシステム。
- 12前記自己相関手段は、一連の出力サンプル{1...N}を有する前記自己相関関数を発生させ、前記ピケットテスト手段は、N/3未満のラグ深さを有する候補ピークに関して少なくとも2つのピケットが存在するかどうか、および、N/3よりも大きくN/2未満のラグ深さを有する候補ピークに関して少なくとも1つのピケットが存在するかどうか、を識別するように構成されている、請求項1乃至11のいずれか1項に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 13前記ピークセレクタ手段は、頻脈ラグ閾値未満のラグ深さの中に位置付けされている頻脈閾値よりも大きい前記自己相関関数の中の任意のピークが存在するかどうかを決定し、そうである場合には、可能性のある頻脈性不整脈に関してフラッグをセットするように構成されている、請求項1乃至12のいずれか1項に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 14前記ピークセレクタ手段は、前記自己相関関数の中のピークのセットの間で2段脈パターンを識別するための2段脈識別手段をさらに含む、請求項1乃至13のいずれか1項に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
- 15前記ピークセレクタ手段は、前記自己相関関数の中のピークのセットの間でジッタパターンを識別するためのジッタ識別手段をさらに含む、請求項1乃至14のいずれか1項に記載の植え込み型の医療用デバイスシステム(12、14、32、34)。
Independent claims15
163 paragraphs, as filed
0001The present invention relates to peak selection for heart rate autocorrelation analysis in implantable medical devices.
0002Implantable defibrillators are designed to deliver electrical stimulation and terminate certain harmful arrhythmias. Such devices must be able to correctly identify dangerous arrhythmias (sensitivity). They must also avoid delivering electrical stimuli when undesired (specificity). Obtaining high sensitivity and specificity in distinguishing such harmful arrhythmias is challenging.
0003Typically, treatable arrhythmias include ventricular fibrillation (VF) and / or polymorphic ventricular tachyarrhythmia (PVT). Other arrhythmias can include monomorphic ventricular tachyarrhythmias (MVT), atrial fibrillation (AF), and atrial flutter (flutter), and AF and flutter atrial arrhythmias are supraventricular. Considered as sexual tachyarrhythmia (SVT). For some patients, MVT is treated with an implantable defibrillator that uses anti-tachycardia pacing (ATP), while AF and flutter are typically treated entirely by other therapies. .. In addition, patients may experience exercise-induced ventricular tachycardia (VT), which is typically not treated at all. Some patients experience bundle branch block and other illnesses that can occur at elevated rates, and the signal shape (morphology) of the cardiac signal with each heartbeat changes relative to morphology at lower rates. Implantable devices are expected to properly distinguish between these various illnesses and to provide the correct treatment for only certain illnesses.
0004In Non-Patent Document 1, Chen et al. Discuss the use of an autocorrelation function (ACF) to identify ventricular fibrillation, and ACF is performed on ventricular fibrillation. Chen that the peaks in the ACF output are expected to be periodic and / or periodic and should pass the linear regression test when ventricular tachycardia (VT) is occurring. (Chen) et al. Assume. Therefore, the result of ACF undergoes a linear regression analysis and a VF is declared if the linear regression cannot find a linear fit. Chen et al. Limit their analysis to VF and VT, and the linear regression they are discussing includes atrial flutter or atrial fibrillation, where defibrillation treatment is typically undesirable. It does not address the tendency to fail with respect to supraventricular arrhythmias, such as. Moreover, adding ACF to linear regression testing creates a huge computer computational burden for implantable systems.
0005Sweeney et al. Discuss in Patent Document 1 and / or Patent Document 2 the use of ACF applied to the conversion of detected cardiac signals using curve matching. ACF will be applied to identify repetitive curves. Such a repetitive curve can be used to find the heartbeat from the transformed signal, which can be used to calculate the rate. However, ACF does not apply directly to time-varying cardiac signals.
0006In each of these examples, the ACF requires that a large number of computer computing steps be to be calculated. In order to make ACF more useful in implantable devices, simplified and alternative methods are desired to address the spectrum of possible arrhythmias.
<p num="0007"><patcit num="1"><text>U.S. Pat. No. 8,409,107</text></patcit><patcit num="2"><text>U.S. Pat. No. 8,521,276</text></patcit></p>
<p num="0008"><nplcit num="1"><text>Ventricular Fibrillation Detection By A Regression Test On The Autocorrelation Function, Med Biol Eng Comput .; 25 (3): 241-9 (May, 1987)</text></nplcit></p>
<p num="0009"> The inventor has recognized, among other things, that the problem to be solved may involve incorporating a modified autocorrelation function into an implantable cardiac device. Modifications can be made to reduce the computer computational burden of correlation functions and to adapt to some of the difficulties that arise in the context of implantable devices that monitor cardiac function. This subject can help provide solutions to the challenges of enhancing sensitivity and specificity within implantable cardiac rhythm management devices.</p>
<p num="0010"> The present invention consists of several separately mountable elements that provide a way for reliable use of ACF within an implantable device with less computer computational burden.</p><p num="0011"> In the first aspect, the invention consists of a set of rules for the calculation of the minimum absolute difference (MAD) function to build an autocorrelation. The use of MAD facilitates a mode of analyzing cardiac signals that is much easier and less computer-intensive than ACF.</p><p num="0012"> In a second aspect, the invention comprises a set of rules for the identification and selection of candidate peaks within an autocorrelation or ACF to produce an estimate of heart rate. In a third aspect, the invention comprises a set of rules for tracking heart rate over time using output peaks from either autocorrelation or ACF.</p><p num="0013"> The first, second, and third aspects can be used independently of the other aspects, respectively, or such as the 1st 1-2nd, 1st-3rd, or 2-3rd aspects. , Can be used in any suitable combination.</p><p num="0014"> In a fourth aspect, the invention comprises an integrated system or method in which the simplified autocorrelation of the first aspect is combined with the second and third aspects. In various embodiments, the device and method can use any of the first to fourth aspects, either continuously or following a trigger event.</p><p num="0015"> The present disclosure relates to enhancements that, in certain cases, assist in selecting peaks in the results of autocorrelation functions. The autocorrelation function can take the form of a simplified ACF, such as MAD, which is, for example, the calculation of autocorrelation in an implantable cardiac device (CALCULATION OF SELF-CORRELATION IN AN). Includes the method of US Provisional Patent Application No. 62 / 038,440 entitled IMPLAN TABLE), but the results of more complex calculations may also be used. CARDIAC RATE TRACKING IN AN IMPLANTABLE MEDICAL in an implantable medical device Tracking analyzes, such as those in US Provisional Patent Application No. 62 / 038,437 entitled DEVICE), use the results of peak selection for therapeutic use and other decisions over time. It is possible to provide rate estimates and the confidence added is based on repeated peak selection. However, another example omits tracking and selects from the available peaks of the autocorrelation function to provide an estimated heart rate.</p><p num="0016"> Peak selection enhancement helps generate rate output from noise and data that can change in response to other inputs, which can complicate the extraction of heart rate from signals detected by implantable systems. Above all, the cardiac signals themselves are variable beyond noise and external inputs, assuming that an implantable system is provided to patients with different heart diseases or abnormalities in different forms and conditions. possible.</p><p num="0017"> This summary is intended to provide a summary of the subject matter of this patent application. This overview is not intended to provide an exclusive or comprehensive description of the invention. Detailed description is included to provide further information about this patent application.</p><p num="0018"> In drawings that are not necessarily drawn to the correct scale, similar numbers can explain similar components in different figures. Similar numbers with different letter suffixes can represent different examples of similar components. The drawings generally illustrate, by way of example, the various embodiments discussed in this document, but not by limitation.</p>
0019<figref num="1">The figure which shows the cardiac treatment system implanted under the skin.</figref><figref num="2">The figure which shows the cardiac treatment system implanted in a transvenous.</figref><figref num="3">The figure which shows the overall method for generating a heart rate estimate.</figref><figref num="4">Diagram that illustrates the analysis of data using autocorrelation functions and differentiates such analysis from ACF.</figref><figref num="5">Diagram that illustrates the analysis of data using autocorrelation functions and differentiates such analysis from ACF.</figref><figref num="6">The figure which illustrates the analysis of the R [n] peak for identifying a candidate heart rate and a heart rate estimate.</figref><figref num="7">The figure which illustrates the analysis of the R [n] peak for identifying a candidate heart rate and a heart rate estimate.</figref><figref num="8A">The figure which illustrates one method of tracking heart rate.</figref><figref num="8B">The figure which illustrates one method of tracking heart rate.</figref><figref num="9">The figure which illustrates the tracking of the heart rate over time.</figref><figref num="10">A diagram demonstrating several heart rate tracking steps using a virtual example.</figref><figref num="11">A diagram demonstrating several heart rate tracking steps using a virtual example.</figref><figref num="12">A diagram demonstrating several heart rate tracking steps using a virtual example.</figref><figref num="13">A diagram demonstrating several heart rate tracking steps using a virtual example.</figref><figref num="14">The figure which illustrates the dominant peak test about the peak selection of FIG.</figref><figref num="15">The figure which illustrates the high rate peak test about the peak selection of FIG.</figref><figref num="16">The figure which illustrates the analysis for identifying the bigeminy pattern and correcting the rate analysis.</figref><figref num="17">The figure which illustrates the analysis for identifying the jitter and correcting the rate analysis.</figref><figref num="18A">Diagram illustrating different scenarios linking R [n] calculations, peak selectors, trackers, and treatment decision blocks together.</figref><figref num="18B">Diagram illustrating different scenarios linking R [n] calculations, peak selectors, trackers, and treatment decision blocks together.</figref><figref num="19">Block flow diagram for the overall method of cardiac signal analysis.</figref>
0020Figures 1 and 2 show implantation sites for an exemplary cardiac system. The present invention may be applied to a dedicated subcutaneous system as shown in FIG. 1 or to a transvenous system as shown in FIG. Alternatives can include systems with multiple subcutaneous, transvenous, and / or intracardiac elements, epicardial systems, or fully intravenous or intracardiac systems.
0021The exemplary system shown in FIG. 1 is shown for heart 10 and is also intended to carry subcutaneous implants that will result on the patient's ribs and under the patient's skin. Has been done. The canister 12 is implanted near the left axilla and can be in a lateral, anterior, or posterior position. Lead 14 connects the canister 12 to electrodes 16, 18, and 20 such that electrodes 16, 18, and 20 are implanted along the patient's sternum, typically to the left or right of it. Is illustrated in. The system of FIG. 1 can include an external programmer 22 that is configured to communicate with the implant 12.
0022The system in Figure 2 is the transvenous system, which is again illustrated for the heart 30 and the patient's ribs are omitted for clarity. The canister 32 is in a high chest position and the lead 34 has access to the vascular system and into the heart. The lead 34 can include a superior vena cava coil electrode 36, a right ventricular coil electrode 38, and one or two ventricular sense / pace electrodes 40, 42. Again, the programmer is configured to communicate with the implanted system, as shown in 44. The system can further include a left ventricular lead (not shown).
0023Communication with respect to either the system in Figure 1 or Figure 2 is guided, RF, direct (ie, using the patient's own tissue as the communication medium), or of any other suitable communication. It can be via a medium. Such communications configure an implanted system for sensing, treatment, or other features, load new software or firmware for the implanted system, and device state, It can be useful for eliciting information about system behavior, such as treatment history, diagnostic data (both device and related patients), or other relevant data. Programmers can include such circuits as needed to provide processing, memory, displays, and telemetry / RF communications, etc., for these stated purposes.
0024The canisters of FIGS. 1 and 2 will typically contain arithmetic circuits for implantable systems. Arithmetic circuits are required for signal processing, memory storage, and the generation of high-power electrical output, low-power electrical output, and / or non-electrical output, as well as optional. Suitable analog and / or digital circuits can be included. For example, analog-to-digital converters (ADCs) are direct conversion ADCs, successive approximation ADCs, ramp comparison ADCs, Wilkinson ADCs, integration ADCs, dual-slope ADCs or multi-slope ADCs, pipeline ADCs, or sigma-delta ADCs. It is possible. Other ADC types, modifications and / or hybrids of any of these types, may be used instead, as will be appreciated by those skilled in the art.
0025The arithmetic circuit can be coupled to the appropriate battery technology for implantable devices by any of a number of well-known examples in the art, and it uses a variety of capacitor technologies for a short period of time. It is possible to support build-up and / or energy storage for defibrillation or other high power purposes. Leads and outer shells for canisters can be manufactured from a variety of materials suitable for implantation, such as those widely known throughout the art, with coatings on such materials. For example, the canister can be made using titanium, with a coating of titanium nitride or iridium oxide (or other material) if desired, and the leads can be made of polyether, polyester, polyamide, polyurethane, or. It can be formed of a polymeric material such as polycarbonate or other material such as silicone rubber. The electrodes can also be similarly formed from a suitable material, such as silver, gold, titanium, or stainless steel such as MP35N stainless steel alloy, or other materials.
0026The location of system implantation can change. For example, the system shown in Figure 1 is a dedicated subcutaneous system located in the precordium and lateral chest between the patient's skin and thorax. Other subcutaneous-only systems (including systems without leads 14, systems with multiple leads 14, or alternative arrays of leads 14) have other anterior-only installations and / or anterior-posterior locations, posterior. Can be used only in place, laterally, etc., for example, in US Pat. No. 6,647,292, US Pat. No. 6,721,597, US Pat. No. 7,149,575, US Pat. No. 7,194,302. Including the locations described, each of which is incorporated herein by reference and may be used elsewhere as well. Subcutaneous placement can include anywhere between the skin and the thorax, including submuscular.
0027In addition to those in Figure 2, exemplary transvenous systems include single-chamber systems, dual-chamber systems, and biventricular systems. Also, a completely intravenous system has been proposed. Additional or other coatings or materials different from those described above may be used, among other things, with respect to epicardial, transvenous, or intravenous systems, leads, and canisters. The system can further include an implantable "seed", which can be attached directly to the myocardium without the use of any leads. Some systems allow the combination of implantable intracardiac seeds and defibrillators for subcutaneous use, where seeds and defibrillators provide controlled treatment and / or perceived data. It is effective for two-way communication such as transportation of.
0028Various alternatives and details regarding these designs, materials, and implantation approaches are known to those of skill in the art. Commercially available, including Boston Scientific Teligen ICD and S-ICD systems, Medtronic Concerto and Virtuoso systems, and St. Jude Medical Promote RF and Current RF systems. Systems are known, in which the above methods may be implemented, or it may be configured to implement such methods. Such platforms include numerous examples and alternatives for various system elements.
0029As illustrated and illustrated, there are various schemes in which an implantable cardiac rhythm management device or system can be implemented with respect to the present invention. In some examples, the methods and devices focused in the present invention include the ability to acquire and analyze long-range cardiac signals. Some examples of long-range signals include signals obtained between two subcutaneously placed electrodes or between a canister electrode and an intracardiac electrode. Short-range signals and / or signals generated as a combination of short-range and long-range signals can be evaluated in other alternatives.
0030Figure 3 shows the overall method for generating heart rate estimates. The function "R [n]" is calculated. The function R can be, for example, an autocorrelation function as shown in FIGS. 4-5 below. The function R represents a series of comparisons made by computing a relatively large number (50 or more) of comparisons between the comparator that is part of the signal and the overall signal itself. , By repeatedly shifting the comparator over the overall signal.
0031R is discussed herein as a distributed function rather than a continuous function. In another example, R can also be a continuous function. In the example, R [n] can be a function that can be called periodically, or it can be generated continuously. Briefly referring to FIG. 5, an exemplary R [n] as calculated at the time of selection is shown in 140 based on a scrolling comparison of the comparator 122 to buffer 120, which is buffer 120. , Has a length M, and the comparator has a length M / 2, providing R [n] with a length of M / 2. It is possible to think of R [n, t], where n has a value that represents the individual calculation of R at time t. For example, Figure 10 shows three "R" functions at t1, t2, and t3, respectively, where R was calculated for n = 0 to 400.
0032By using R [n] as calculated from step 60, some candidate peaks are identified at 62. 6 to 7 provide an example of identifying candidate peaks. Candidate peaks can be understood as representing possible "rates" of cardiac events. High matching, as represented by the peak of R [n], suggests matching of the periodic electrical waveforms associated with the heartbeat. For example, if the peak of R [n] occurs at n = 90 and the sampling rate is 256 Hz, then the time between R [0] and the peak will be 90/256 = 352 ms. For this example, shifting the comparator time back by 352 ms produces a relatively high match between the comparator and the original signal. 352ms can be referred to as the lag depth, and if it is truly the distance between consecutive R waves, it corresponds to 171 beats per minute (bpm). Become.
0033The method then determines at 64 whether a valid track exists. Tracking is the process of monitoring the output of an R [n] calculation and the peaks from it to determine if a highly reliable heart rate can be reported. 8A-8B show an exemplary method of rate tracking. Figure 9 shows another approach to rate tracking.
0034If the track already exists, the method involves deciding whether to identify one of the candidate peaks, as shown in 66. If no track exists, peak tracking is performed as shown in 68 to determine if a new valid track can be declared. Then, following either 66 or 68, the method ends the iterative calculation shown by reporting the rate and confidence.
0035In some cases, high confidence rates will not be reported. For example, atrial arrhythmias conducted in the ventricles can be characterized by a period of instability between ventricular depolarizations. As a result, the measured ventricular rate can be very variable. When the ventricular rate is very variable, R [n] can produce only relatively low peaks or cannot consistently produce similar peaks during iterative calculations. As a result, the output rate from the entire procedure is either missing or simply reported in block 70 with low confidence. 10 to 13 illustrate examples showing some iterative calculations of the method of FIG.
0036In the example, the output rate and confidence can be used to confirm or question the heart rate as calculated using more conventional processing. For example, the device can use a default pulsation detection scheme, in which the cardiac signal received following amplification and filtering is compared to the detection threshold. Several exemplary pulsatile detection approaches are set forth, for example, in US Pat. No. 8,565,878 and US Pat. No. 5,709,215, the disclosure of which is incorporated herein by reference. Exceeding the detection threshold can then be presumed to represent a pulsation or R wave, and various known methods are used to identify and eliminate exceeding the detection threshold caused by noise or overdetection. Can be done. See, for example, US Pat. No. 7,248,921, US Pat. No. 8,160,686, US Pat. No. 8,160,687, US Pat. No. 8,265,737, and US Pat. No. 8,744,555. Those disclosures are incorporated herein by reference.
0037The remaining detected beats or R waves, and the intervals between them, can be used to calculate the rate. In some embodiments, the present invention is used to double-check rates calculated using such methods. Such double checks may be requested or continuously provided as needed. For example, a double check can be performed to confirm the rate before treatment is given or before preparation for treatment is given. In some embodiments, the invention is capable of providing rate estimates that can overturn rates as calculated using other methods such as pulsation detection.
0038In another example, the double check can be performed to confirm accurate event detection as a method of verifying the sensing configuration, where the sensing vector can be changed if the sensing configuration has not been verified. In other embodiments, the elements of the invention can be used to provide rate calculations by default, or can be the sole source of rate calculations.
0039Figures 4-5 illustrate the analysis of data using autocorrelation functions, which differentiates such analysis from ACF. Specific additional options and examples can be found in US Provisional Patent Application No. 62 / 038,440 entitled CALCULATION OF SELF-CORRELATION IN AN IMPLANTABLE CARDIAC DEVICE. Can be issued.
0040FIG. 4 shows the sensed ECG signal at 100. The signal can be treated as a buffer of length M, as shown in 102. The calculations for R [1], R [2] ... R [M-1] are illustrated in 104. Each calculation of R [n] is performed in a normal ACF by multiplying a portion of the buffer (via the dot product) by a portion of the comparator, with the comparator time shifted relative to the buffer. Will be done. The comparator itself is simply a copy of the original buffer. Due to the time shift, a correction factor is required as shown in 106. The reason is that the dot product is calculated using data points that are getting smaller and smaller as the size overlap is reduced with each successive calculation of R [n]. Shifting the comparator can be referred to as lag depth.
0041The first simplification is to trade multiplication for subtraction to calculate the dot product. The absolute value of the subtraction result produces the minimum absolute difference (MAD). Replacing the dot product and using MAD instead would reduce the number of calculations required by more than an order of magnitude, with minimal reduction in accuracy.
0042The buffer 110 with length M is then split in half to provide sample portion M / 2 and available lag depth 114 to eliminate the correction factor 106 for duplication. The iterative comparison then identifies the area of difference between sample 112 and buffer 110. As shown in 116, the result is an M / 2 overall comparison from zero lag depth to M / 2 lag depth.
0043Additional simplification can be accomplished by compressing the input data in some embodiments. For example, the system can perform analog-to-digital conversion of cardiac signals at a rate of 256 Hz. The calculation of R [n] can be performed on a limited or compressed version of the original signal, reducing the number of calculations again (spends the calculations required for downsampling, but it is data storage. Can already be done to facilitate).
0044Looking at Figure 5, an exemplary calculation of R [n] at a particular point in time is shown. ECGs such as those stored by buffers of length M are shown at 120. A comparator for autocorrelation is shown in 122, which contains half of the buffer 120 with the most recently detected samples. Preferably, the length of M is sufficient so that at least two beats fit into the comparator during a benign rate (eg 60 bpm). Thus, in the example for illustrative purposes, the buffer 120 has a length of about 4 seconds and the comparator 122 has a length of about 2 seconds. Another example has a buffer 120 that is about 2 seconds long and the comparator 122 is about 1 second long. Other sizes may be used. In some examples, the invention is characterized by having a buffer large enough to ensure that at least two cardiac cycles that occur at the required lowest rate are acquired. , The lowest required rate can be in the range of 60-120 bpm. In a further illustration, the buffer length can be between 1.5 and 6 seconds, and the comparator length is between 750 ms and 3 seconds. In the example shown herein, the comparator is half the length of the buffer, and in other examples, the comparator is between one-tenth and one-half of the total buffer length. It is possible that there is.
0045As shown in 124, the MAD function is applied in this example and then normalized using the maximum of MAD over all comparisons made for a particular iterative calculation of R [n]. This produces the results graphed in 126. The resulting graph contains a peak at 130, which corresponds to the zero lag depth calculation, in which the MAD becomes zero, giving an output of 1. The following peaks 132, 134, and 136 correspond to the time when the MAD is calculated while the R wave peak in the comparator is matched with the set of R wave peaks from the buffer, respectively. For example, if peak 138 of the comparator is matched with peak 140 of the buffer and the R wave spacing is similar, this will also match adjacent peaks, giving a small absolute difference in that particular match. It will be. In the analysis, zero lag depth calculations are typically ignored.
0046Other peak matching between the comparator 122 and the buffer 120 can produce smaller peaks at R [n]. For example, peak 142 occurs when the comparator peak 138 is matched with the T wave at 144. This positioning produces a smaller MAD output, which, when normalized using formula 124, produces an easily noticeable but small peak in R [n].
0047The R [n] function can be calculated periodically. In one example, the buffer and comparator take a fairly large constant time, for example greater than 1 second, or even 2 seconds, so the need to continuously recalculate R [n]. Does not exist. For example, the period between recalculations of R [n] can be approximately the duration of the comparator, or, in another example, approximately half the duration of the comparator. For example, if the comparator length is 2 seconds, the buffer can be 4 seconds long and the calculation of R [n] can be performed at 1 second intervals. Therefore, every second, the buffer will be updated, the comparator will be reformed, the sequence of comparisons, and the time shift will be repeated. For example, Figure 10 shows a repetitive calculation of R [n] at t1, t2, and t3, so R [n, t1], R [n, t2], and R [n, t3] , Shown in 200 of that figure.
0048In the present specification, examples for illustration purposes in various figures propose the use of asynchronous calculation of R [n]. These are asynchronous unless the calculations are linked or synchronized with the beat detection performed by some other method. In other embodiments, it is possible to use the update or recalculation of the pulsatile synchronous expression of R [n] instead. The hybrid embodiment can be updated synchronously to take advantage of the microprocessor / microcontroller wakeup caused by pulsation detection, but the calculation of R [n] is the same as some desired metrics. It can be limited to occur as often as possible. For example, the calculation of R [n] can be a beat that is synchronized at intervals of 1 second or longer.
0049The simplification of the optional choices in FIGS. 4-5 are provided for illustration and illustration purposes. However, some of these simplifications are because the peak selection and / or tracking examples shown below do not depend on any particular type of calculation for R [n] unless otherwise stated. In the embodiment of, it may be omitted.
00506-7 illustrate the analysis of R [n] peaks to identify candidate heart rate and heart rate estimates. FIG. 6 shows the operation in a flow diagram, while FIG. 7 provides a graphic example.
0051In FIG. 6, at the beginning at block 150, the method begins with the identification of any peak in R [n, tk]. In this example, peaks that are within 50% of the maximum peak and meet some minimum size criteria are reported to the peak tracker, as shown in 152. Tracking can be performed, for example, as shown below in FIGS. 8A-8B and 9.
0052The maximum candidate set is then selected as shown in 154. In the example for exemplary purposes, the threshold can be set in the scaled calculation of R. For example, using formula 124 in Figure 5, the threshold for candidate peaks can be set to R = 0.3, so that the peak must be greater than 0.3 times the maximum peak to be considered as a candidate. It has become. The maximum peak will always occur at R [0]. The reason is that it is when the comparators and buffers are perfectly matched, and therefore when R [0] = 1. Any subsequent peak greater than 0.3 can be a candidate peak. In the example, up to 5 maximum peaks greater than 0.3 (excluding peaks at R [0]) are treated as candidates.
0053As shown in 156, the tachycardia flag is set if any of the largest candidates has a lag depth that would place the candidate within the "tachycardia zone". If the peak lag depth is relatively small, the candidate peak is in the "tachycardia zone". This can be identified by querying for the presence of a peak at R [nt], where nt is less than the tachycardia threshold. For example, the tachycardia zone flag should be set for candidate peaks suggesting rates above 160 bpm, and if the sampling rate is 256 Hz, then the peak at n <96 is in the tachycardia zone. The reason is that the peak will occur at a lag depth of less than 375 ms, which is equal to a rate above 160 bpm. When the tachycardia flag is set, the tachycardia flag indicates that the analysis suggests the possibility of tachycardia, even though the rate at which the analysis finally concludes is generally correct. There is.
0054The first candidate peak is then selected as shown in 158. One of two rules for finding candidate peaks can be applied. That is, a peak with a lag depth that allows a rate greater than 75 bpm to be found, which is greater than the first peak in time by the selected critical delta, is shown at 160. Or otherwise, the first temporal peak of the candidate peaks is chosen as shown in 162. The first peak in time is the candidate peak with the smallest lag depth. The first rule at 160 allows a peak that is significantly larger than the first candidate peak in time to be selected as long as it exceeds the rate threshold. In the example, the rate threshold is 75 bpm and other thresholds may be used.
0055In the example for illustrative purposes, the combination of 160 and 162 ensures that the peaks associated with the higher rates are analyzed first, and in this regard, the method is attached to look for higher rate candidates. Momentum. Energizing towards higher rates may be desirable to minimize the risk of heart rate underestimation in the presence of tachyarrhythmias.
0056Candidate peaks are then analyzed by looking for and counting "pickets", as shown in 164. Pickets are peaks at multiple lag depths of candidate peaks. FIG. 7 shows an example of a picket. The first peak is found at a lag depth of 110 samples (corresponding to 140 bpm sampled at 256 Hz). This lag depth gives an RR interval as shown at 180. The two pickets can be identified by observing additional peaks at 220 and 330 sample lug depths, where the 220 and 330 sample lag depths are multiple candidate lug depths. Pickets at 182 and 184 provide confirmation that the 140bpm rate is likely to be the correct heart rate.
0057Counting pickets can include tolerances for some changes in peak spacing. For example, picket peaks should be evenly spaced within maximum tolerances. Tolerances can be defined as a function of calculated heart rate, or can be set in milliseconds or in terms of sample (n). For example, if the first peak is in a lag of 80 samples (313 ms at 256 Hz), pickets may appear between 75 and 85 samples (293 ms to 332 ms) apart. It will be expected. Narrower or wider tolerances can be defined in other examples.
0058Note that in this example, the maximum peak is not the first selected candidate peak. There are two reasons why this is so. First, the maximum peak is not large enough for the candidate peak to meet Rule 160. In the example, in order to select a peak other than the first peak as a candidate, subsequent peaks need to be at least 30% larger than the candidate (creating a delta relative value), which is the case here. Does not apply to. In another example, the delta can be a fixed value, for example 0.2, using the MAD formula 124 from Figure 5.
0059Second, the maximum peak is at the lag depth corresponding to the rate of 70 bpm, which again does not meet Rule 160. In the example, in order to select a peak other than the first peak as a candidate, subsequent peaks are required to be at a lag depth corresponding to a rate greater than 75 bpm. Other thresholds may be chosen. The picket determination is that the peaks used to establish subsequent pickets are between the maximum N peaks, or each peak is 0.35 or 0.50, for example, using the formula in 124 in Figure 5. It is possible to request that it be greater than a predetermined threshold, such as a peak that exceeds it.
0060Both rules 160 and 162 can be modified in other examples. For illustrative purposes, FIG. 7 also illustrates tachycardia zone 186. In this example, the tachycardia zone covers a lag depth from zero to about 90. This corresponds to offsets of up to 90 samples. In the illustration shown, 384 samples equals 1.5 seconds, which means a 4ms sampling period. 90 samples will correspond to an estimated RR interval of 360ms, which is equal to 167bpm. As mentioned above, other settings for the tachycardia zone may also be used.
0061Returning to FIG. 6, the method proceeds by determining whether the picket test has passed, as shown in 168. In the example, the picket test passes if there are at least two pickets identified for the candidate peak. In another example, multiple picket thresholds may be applied, depending on the lag depth of the peak under the analysis. For example, in an analysis with a maximum lag depth of N, the set of rules requires at least two pickets for candidate peaks with a lag depth of less than N / 3, and is greater than N / 3 and less than N / 2. It is possible to request at least one picket for a candidate peak with a lag depth of. This relative approach is simply 1 because for candidate peaks with a lag depth between N / 3 and N / 2, the second picket will be at a greater lag depth than the "N" itself. Address that only one picket is possible. Such relativity is optional and can be at least partially controlled using the dominant peak test discussed below.
0062If the picket test passes, the method makes a final check for any large peak in the tachycardia zone, as shown in 170. In some limited cases, numerous peaks can be reported during chaotic tachycardia events. In such cases, the decision to select only the "N" maximum candidate peak in block 154 cannot select the peak in the tachycardia zone as a candidate. Therefore, the test at 170 looks for any peak in the tachycardia zone that is within 30% of the size of the maximum peak but has not been identified as a candidate.
0063If a large tachycardia zone peak is identified at 170, this peak also undergoes a picket test. If the peak selected at 170 has a picket that passes the picket test, a large tachycardia zone peak from block 170 is reported as an RR estimate at 174. Otherwise, the tachycardia zone flag will be set and candidate peaks that pass the picket test at 168 will be reported as RR estimates at 174, as described in 156.
0064In some examples, all candidates can be checked until one is found that passes the picket test, and if no picket test pass is found for any of the candidate peaks, the method is Proceed to block 172. Alternatively, if only the first selected candidate peak takes the picket test at 168 and does not pass the picket test once, the method goes to block 172.
0065Upon reaching block 172, the dominant peak test is applied. The dominant peak test determines if there is a peak in R [n] that is 30% larger than all other peaks (excluding the peak at n = 0). If so, the dominant peak is identified as an RR estimate.
0066Also, the dominant peak test 172 may be limited to pass when the identified dominant peak is at a lag depth corresponding to a rate below a preset threshold, such as 60, 75, or 90 bpm. The rate limit can be included in block 172 as a confirmation that pickets may not be present in the analyzed data for low rate peaks. This is because the time span of the R [n] calculation may not be sufficient to produce a picket pattern for all heart rates, especially for lower heart rates with longer beat intervals. It has become.
0067For example, using a 3 second buffer and a 1.5 second length comparator, the first picket for a peak at a lag depth of 800 ms (75 bpm) would be at 1.6 seconds. Given a buffer / comparator size of 3 seconds / 1.5 seconds, the maximum lag depth is simply 1.5 seconds, so such pickets cannot be identified given a buffer / comparator size. On the other hand, dominant peaks at smaller lag depths, such as 500 ms (120 bpm), would be expected to have two pickets in this scenario, and the absence of pickets is reliable. There appears to be less periodicity than what would normally be associated with a high RR estimate, so it would not be treated as a likely RR estimate.
0068The RR estimate goes through one of three possible paths: a candidate peak that passes the picket test (158-164-168), a large tachycardia peak (170), or a dominant peak test (172). RR estimates may be reported when calculated. A reliability grade may also be applied. In the example, three grades are available.
0069--High reliability in any of the following cases Rate> Tachycardia threshold with 3 pickets, and R> HC threshold, or Rate <Tachycardia threshold with two pickets and R> HC threshold --Intermediate reliability by default when high or low reliability conditions are not met, and --Low reliability in the following cases R <LC threshold or One or less pickets and dominant peak test 172 have not passed In this example, the tachycardia threshold can be set to a range requiring high rates, for example above 150bpm, 180bpm, or 200bpm. In some examples, the tachycardia threshold can be selected with reference to the buffer size and comparator size to link to the amount of pickets that can appear. The definition of the HC threshold will simply depend on how R was computer-calculated. In the example, the HC threshold is set to 0.65, assuming a computer calculation of R using the formula in 124 in Figure 5. Similarly, the LC threshold will be defined in a manner that is closely tied to R's computer computation. Also, in the example using formula 124 from FIG. 5, the LC threshold is set to 0.35.
0070The reliability information can also be incorporated into the tracking process shown in FIGS. 8A-8B. For example, switching from one track to another, or declaring a new track, can be accelerated in response to a high confidence rate, while such a process is of low confidence. May be delayed in terms of rate.
0071In the two-stage pulse pattern, there are two alternating morphologies for ventricular depolarization or "R" waves. When the two-stage pulse pattern is analyzed using autocorrelation, the R and T wave peaks alternate, have different morphologies, and the output is that R and T wave peaks. It can be difficult to determine whether it reflects two R waves with a two-step pulse pattern. Specific approaches for identifying two-stage pulse and / or jitter are shown below in Figure 16 (two-stage pulse) and Figure 17 (jitter).
00728A-8B illustrate one method of tracking heart rate. Starting from block 200 in FIG. 8A, any suitable mode of finding R [n], selecting a set of peaks, and generating RR estimates is implemented. The methods illustrated in FIGS. 4-7 offer various options for the block 200. The tracking method begins by determining if an existing track or "active" track exists, such as 202. If so, the method goes to B in Figure 8B.
0073If no existing track exists, the method determines if a valid RR estimate has been generated, as shown in 204. If no valid RR estimate has been obtained from the previous analysis, no new track will be declared and the method will end at 212 without a track and wait for the next iterative calculation. If a valid RR estimate is found, then the method is X out of the last Y RR Estimates (or X out of the last Y RR Estimates), as shown in 206. , Attempt) to determine if they are similar. For example, if the RR estimates for the last three quarters are similar, then the test at 206 would be satisfied for 3/4 of the X / Y. In one example, the three-sixth rule applies in 206. If the test at block 206 is met, a new track will be established at 208.
0074If the test at 206 is not met, the new track can still be established based on a single very reliable rate calculation. The definition of very high reliability can change. In one example, the high / intermediate / low confidence rules applied above can be used, and any RR estimate calculated by the high confidence is sufficient to satisfy the rule at 210. Become. In another example, separate thresholds may be set for the very high confidence rules at 210. In one embodiment, in a system that calculates R using the formula shown in 124 of FIG. 5, block 210 is satisfied when R> 0.85. If the rules in 210 are met, a new track will be established as shown in 208. Otherwise, no new track will be set as described in 212.
0075Looking at Figure 8B, it is determined that there are valid tracks and whether recent RR estimates are in the gate, as shown in 220. The gate has a width, the width of which can be defined in various ways. For example, the gate can be 40 milliseconds wide, or it can be 20 bpm wide. The gate can be centered on the previous RR estimate or the average of 2-4 previous RR estimates. In one example, the gate is calculated by converting the latest RR estimate to bpm and by setting the upper and lower boundaries 10 bpm apart. So, for example, if the latest RR estimate is 400ms (which translates to 150bpm), the gate would be 140bpm to 160bpm, and the RR estimate between 429ms and 375ms would be It will be considered as being "inside" the gate.
0076Also, the gate width can take into account rate volatility. For example, the volatility of a recent set of RR estimates can be calculated simply by tracking how much change exists from one estimate to the next. The gate width can be increased if the rate appears to be highly variable in this example.
0077If the RR estimate is inside the gate, the method declares at 222 that the track will continue. Also, RR estimates are reported. If the RR estimate is not inside the gate, coasting rules apply. Coasting occurs when a valid track has been identified / defined, but the iterative calculation of the RR estimate calculation does not produce a result that meets the track definition. The use of coasting allows the track to continue and does not take into account temporary disturbances such as noise or PVC. Coasting avoids gaps in the output RR estimates by retaining the last known RR estimates. When the peak is in R [n], but for some reason otherwise cannot pass the rigorous test in the peak selector to identify the candidate peak as an RR estimate. Coasting can be particularly useful. Coasting is available to rescue RR estimates for such peaks, but only for a limited amount of time.
0078In the example for exemplary purposes, coasting is not allowed to continue forever and limits apply, as shown in 224. If the coasting is within its limits, the method continues through block 232 and at 222, continues the track. Various rules can be applied to limit the coasting, each of which can have different limits.
0079For example, if the RR estimate or peak is not reported from the R calculation, as shown in 226, then there can be a first "no data" limit on the duration of the coasting. Is. In the example, the system would only allow a single iterative calculation of "no data" before declaring track loss at 236. For example, such a "no data" condition can occur if noise interrupts sensing and none of the peaks of R [n] exceed the base threshold. Also, a "no data" condition can occur when a highly polymorphic arrhythmia begins and R [n] simply cannot have any significant peaks. is there.
0080Then, as shown in 228, there is a coasting condition in which no RR estimate is produced, in which all of the reported peaks are gated, as shown in 230. It also demands that it does not exist in. Thus, block 228 covers one set of situations where the rate estimate is missing and the track is unidentified, while block 230 is the track, whether off-track RR estimates have been identified. Covers the situation where a lower confidence check of is present.
0081A fourth form of coasting can occur as part of a transition or "jump" to an alternative track, as described in 232. In this case, by having a "new" track declared using a decision similar to that of 206 or 210 in Figure 8A, the existing track will continue until either alternative track condition is met. ..
0082As can be seen from blocks 226, 228, 230, 232, there are different inputs for the coasting state, each with somewhat different confidence levels. For example, when no data is received 226, the underlying track or sensing reliability is unreliable, and there are no reported RR estimates, and both the reported peaks are at gate 228. Even when it doesn't go inside, it's not so good. These two blocks 226, 228 can be combined with respect to a single limit within the range of 1-3 iterative calculations before the coasting limit is exceeded. Alternatively, block 226 can have a lower coasting limit (1 to 2 iterative calculations), whereas block 228 has an equal or higher limit (1 to 4 iterative calculations). And the combined limits match the higher limits (1-4).
0083An alternative peak in the gate condition is a much higher reliability condition at 230, by suggesting that the track can still be valid, even in the presence of sensing anomalies such as noise. .. This condition 230 can have a higher coasting limit in the range of 2-10 iterations, or simply the overall limit (in the range of 2-10 iterations). It is possible to receive, which will combine any coasting in any of 226, 228, 230, or 232.
0084Block 232 (Jump Limit) allows for a quick transition to a new track without having to first wait for a declaration that the track was lost at 236 before assessing if a new track exists. Exists in. Jump limits also prevent a shift to a new rate based on peaks that will appear in old but no longer valid tracks. When the coasting limit is met at 224 through the jump limit, the outcome follows a different path up to 238. As a result, if the jump limit is met, the method simply continues with the new track definition. In an example for example purposes, the same rules applied in 206 may be applied to the new track to meet the jump limit 232. As described in 232, jump limits can simply be applied to high rate conditions, which are generally of greater concern than low rate conditions, for example in the range 100-180 bpm. Using the limits, above all, 150bpm is one exemplary rate.
0085The use of jumps allows for a quick transition to higher rate RR estimates, and a less stringent set of rules occurs in the presence of existing tracks and jumps. In the example, declaring a new track when the track is not identified would require more confidence in the new data than what is needed for the jump.
0086If the coasting limit is exceeded at 224, the track will be declared lost as described at 236. If the coasting limit is not exceeded, a coasting "state" can be recorded 234, and different coast states are identified for each of the different coasting conditions 226, 228, 230, 232. During the coasting, the track continues as shown in 222 until the next iterative calculation is called.
0087FIG. 9 illustrates heart rate tracking over time. As described in 240, it is possible that there are various triggers for performing autocorrelation. For example, autocorrelation can be the default and continues the analysis required by the implantable system throughout the life of the system. Alternatively, autocorrelation can be required in response to potentially identifiable conditions that require treatment, such as elevated rate conditions. In one example, heart rate can be calculated using conventional R-wave detection schemes (often by comparing the detected heart signal with a time-varying threshold). If the identified rate exceeds the threshold, the autocorrelation method can be initiated to confirm the elevated rate. The threshold can be set, for example, in the range of 100 to 180 bpm, or higher or lower, as desired.
0088In one example, the cardiac treatment system uses a number of intervals to detect (NID) approach or an X / Y filter to transition from apathy to treatment preparation and treatment. It is possible to do. For example, 18 of 24 may be required before the X / Y filter analyzes the detected heartbeats and considers them treatable before treatment is performed. For such systems, if X reaches a lower threshold, eg 8/24, the autocorrelation analyzes and confirms the calculated rate before the 18/24 boundary is reached ( Or it can be called to start rejecting). Similarly, when the NID approach is used, a NID threshold below the therapeutic boundary can be used to trigger an autocorrelation analysis.
0089In another example, autocorrelation can be called to periodically confirm sensing integrity by calculating heart rate for comparison with other rate calculation methods / circuits. In some examples, the autocorrelation presented in this application can serve as the sole estimator of heart rate within an implantable device.
0090When the analysis is triggered at 240, autocorrelation is performed at intervals and R [n, t] is t = {0, 1, .. as shown at 242, 244, 246. It is designed to be calculated for each of the .i}. From this series of calculations, a rate track is sought and, if possible, established as shown in 248. The analysis can confirm or reject the calculated rate, as shown in 250.
0091In addition, the analysis can be used to confirm, accelerate, or delay treatment delivery, as described in 252. Returning to the above example, when the heart rate identified by conventional R-wave detection exceeds a threshold, the autocorrelation is called when the autocorrelation confirms an elevated heart rate in need of treatment. , The treatment threshold can be lowered. For example, if the system uses an X / Y counter set to 18/24, the counter will check for a very high autocorrelation rate before the X / Y counter condition is met. Can be reduced to 12/16. In another implementation, the autocorrelation RR estimate is traditionally calculated for a particular time period, for the amount of events detected, or until the next calculation of R [n] and its analysis. It is possible to exchange heart rates.
0092Figures 10 to 13 demonstrate several heart rate tracking steps using virtual examples. Figure 10 illustrates the start of rate tracking activity. The autocorrelation function is calculated at time t1, t2, and t3, respectively, as shown at 260, 262, and 264. For the purpose of understanding the behavior of this embodiment, the graph in 266 illustrates how the peaks of each R [n] calculation align with each other. Looking at R [n, t1], the graph at 260 illustrates that three peaks above the R = 0.3 threshold were found at approximately 95, 190, and 285 sample lag depths. Using the method of Figure 6, these three peaks 268, 270, 272 will be reported from the peak analysis.
0093Then, using the rule shown in FIG. 6 again, the first peak 268 from R [n, t1] is chosen as the candidate peak. Then the picket will be sought. As shown, for the additional peaks at 270 and 272, there are two pickets identified for candidate peak 268. Therefore, the method of FIG. 6 confirms that peak 268 provides an RR estimate, although peaks 268, 270, and 272 will each be reported to the tracking engine.
0094The RR estimate from R [n, t1] is shown in graph 266 at 274, and other peaks from R [n, t1] are also shown as alternative peaks. Similarly, RR estimates result from the analysis of two other calculations in R [n, t2] and R [n, t3], as shown in 276 and 278. Here, the track has not yet been declared. As a result, each of the RR estimates can be considered to produce an intermediate confidence rate estimate until the track can be declared.
0095Looking at Figure 11, the matching of the results for each of R [n, t1], R [n, t2], and R [n, t3] is the track in Figure 8B using the 3/6 rule. Enough to meet the definition. Therefore, the track gate is shown at 280 for use in evaluating the next iterative calculation of the autocorrelation at R [n, t4]. The newly calculated R [n, t4] is shown graphically in 282. In R [n, t4], in addition to the much higher peak for the R wave at 286, a peak associated with the T wave comparison appears at 284. Applying the set of rules of FIG. 6 again, the first peak is at 284 and can be chosen if the rules at 162 in FIG. 6 are controlled. However, the second peak at 286 appears at a lag depth that is significantly larger than the first peak and supports rates greater than 75 bpm, satisfying the rule at 160 in Figure 6. Therefore, peak 286 was selected for analysis and, as before, was found to have two pickets (not shown) and is used to report RR estimates. As shown in 288, the RR estimate for R [n, t4] is in the gate.
0096Also note that the peak 286 used for the RR estimate exceeds the HC threshold for high reliability. Therefore, the RR estimate at 288 will then be used with high confidence for rate reporting by trackers.
0097Figure 12 shows a different scenario for the calculation of R [n, t4] after the track has been established and the gate has been set at 290. Here, the output changed dramatically from R [n, t3] to R [n, t4], as shown in 292. Using the rule of Figure 6, the first peak 294 is chosen as the candidate peak, but the next significant peak at 296 is too far away so no picket is found. Also, there is no dominant peak. As a result, RR estimates are not calculated.
0098Looking at the updated overall graph, we can see that Gate 298 is empty and has no RR estimates or alternative peaks in it. As a result, for R [n, t4], the analysis is in the coasting state in FIG. Since the RR estimate cannot be calculated based on R [n, t4], the rate will not be reported to the peak tracker. Since the track continues in the coasting state, an output rate estimate will be provided, which in the example would be the value in the gate, or could be the same as the previous output. is there. Based on the empty gate at 298, the output rate estimate will be given a low confidence level.
0099Figure 13 represents another different scenario for the calculation of R [n, t4] after the track has been established. Gate 300 is set up for analysis, but R [n, t4] has a peak located within the gate, as shown at 302, as shown at 304. Not identified as an RR estimate. Instead, peak 306 at a smaller lag depth is identified as the RR estimate. Therefore, one of the alternative peaks at R [n, t4] is in the gate, but the RR estimate 308 is located away from the gate 304.
0100Looking back at Figure 8B, the event shown in Figure 13 would trigger a coasting analysis using the tachycardia jump limit 232. Specifically, the RR estimate 308 is at a relatively short lag depth, which is shown in the example as corresponding to a rate between 180bpm and 240bpm. Given that this is the first such RR estimate, not enough information is yet available to declare a new track. This can be an instant jump, or it can be the beginning of a new rhythm. Until more data is received, the exemplary method will be waiting and coasting with the track continuing. The output rate estimate will continue as it is in the existing track. However, since the RR estimate is outside the gate, any rate estimate will be reported with low confidence. The tachycardia flag will be set based on the large peak in the tachycardia zone.
0101FIG. 14 illustrates a dominant peak test for peak selection in FIG. In the example for illustrative purposes, the results for the R [n, t] calculation are shown with a large peak at 320 with a lag depth of about 200 samples. Here, the signal emits difficulty. First of all, the peak at 320 lacks any picket as shown at 322. The main reason is that due to the large lag depth of peak 320, any peak will appear beyond the end of the R [n, t] calculation. Therefore, in the analytical purpose analysis, the picket test will not pass.
0102Using the analysis of Figure 6, the next candidate peak can be reviewed, where candidate peak 324 can be checked. But again, as shown in 326, the picket will not be identified. There are no picket peaks in the set of identified peaks above the threshold 328 where the pickets must appear in order to pass the picket test for peak 324. In an alternative approach, the single-pass system would only look at a single candidate to identify the picket, and in this analysis, if peak 320 did not pass, the other peaks would not be analyzed. Become.
0103As shown in the lower part of Figure 14, a method for identifying dominant peaks can be initiated by determining that the peak does not pass the picket test, as shown in 340. is there. Then, as shown in 342, it is determined whether there is a peak that is larger than all other peaks (except the null peak at zero lag depth) by some margin. In the example for example purposes, the margin is a percentage X, which can be in the range of 30%, and the range for example purposes is 15% to 50%, or greater or lesser than that. .. Those skilled in the art can define other "margins", for example, depending on the mode in which R [n, t] is normalized, without modifying the principle of the dominant peak test shown in FIG. You will recognize that.
0104Here, the next maximum peak in the R [n, t] graph at 324 is lower than the peak 320 by a margin 330, so that the peak 320 passes the block 342. The dominant peak test then looks to see if the lag depth of the large peak exceeds the dominant peak lag threshold, as shown in 344. The threshold is shown at 332. In the example, the threshold 332 is any peak that does not necessarily pass the picket test due to its lag depth so that the picket cannot appear in R [n, t]. Can be selected to pass. Thus, in the example, within a window with a total depth of 400, the threshold 332 is set at a lag depth of 200. Any peak, such as peak 320, which has a lag depth greater than the threshold 332, will have no picket in the analysis window.
0105Since the peak 320 passes both tests 342 and 344, the peak 320 is identified as a dominant peak and the method would report the rate corresponding to the lag depth of the peak 320. If either test 342 or 344 fails, the method ends at 350 without identification of the dominant peak. As shown in 340, terminating in block 350, assuming the picket test did not pass, in some examples estimates the rate based on a particular R [n, t] calculation. It may result in not being done. In other examples, for example, the maximum peak (320 in this case) can still be used to make low confidence rate estimates. In another example, if no peak is found that passes the picket test, then two possible rates, namely the rate corresponding to the maximum peak 320 and the rate corresponding to the next maximum peak 324, are Each can be reported with low reliability.
0106FIG. 15 illustrates a high rate peak test for peak selection in FIG. Here, many peaks appear in R [t, n], and the maximum five peaks are labeled A, B, C, D, E. Using the picket test rule, the peak at 360 has pickets 362 (corresponding to peak C) and 364 (corresponding to peak E) and passes the picket test. However, another peak 366 was present, which was not selected as one of the candidate peaks due to its somewhat smaller size.
0107The high rate peak test begins by determining whether the picket test has passed, as shown in 370. Here, the picket test has passed as shown by pickets 362 and 364 for peak 360.
0108The high rate peak test then checks for the presence of tachycardia peaks in some percentage of the maximum, as shown in 372. By "tachycardia peak", the method shows a peak that falls within the tachycardia zone and is not between candidate peaks A, B, C, D, and E. In an example for illustrative purposes, using a sampling rate of 256 Hz, the tachycardia zone can be defined as any peak at a lag depth of less than 100, which is for periods of 396 ms or less, and rates above 150 bpm. Will correlate with. In an alternative example, block 372 can identify any peak within the tachycardia zone without reference to the height of the maximum peak.
0109In this example, such a peak appears at 366. The picket test is then re-run using the peak identified in block 372. The retest is shown as an illustration in 380. Peak 366 has pickets 382 and 384 as shown.
0110If the picket test is passed at 374, the first selected peak is replaced by a tachycardia peak identified at 372 and passed the picket test at 374. In the example for illustrative purposes, rather than selecting peak A at 360, the method instead selects peak 366. In this example, the calculated lag depth goes from 120 to 60 and causes an increase in the identified heart rate from 130bpm to 260bpm. In some examples, outcomes can be treated as having lower confidence, or because the peak that produces the heart rate estimate is the lower peak and was not identified in the first pass. , The ambiguity flag can be set to indicate that some ambiguity is present.
0111As mentioned above, in the two-stage pulse pattern, there are two alternating morphologies for ventricular depolarization or "R" waves. The two-stage pulse pattern of ABAB will produce alternating peaks in the autocorrelation result R [n]. When "AB" is compared to "AB", high peaks will appear, each of the A peaks will be matched, and each of the B peaks will be matched, and "AB" A relatively low peak will appear when "" is compared to "BA", that is, when A is compared to B and B is compared to A. However, in the two-stage pulse pattern, the AB interval and the BA interval often match.
0112A cardiac signal with a relatively large T wave may appear somewhat similar to some two-stage pulse signals when compared to an R wave. This requires two elements. That is, firstly, the R and T waves must be very similar overall, and secondly, the RT and TR intervals must also be very similar. The R wave is usually narrower than the T wave, but if both are single-phase, the two can be very similar. The RT and TR intervals are generally similar, only in a narrow range for any given patient. Moreover, as stated, for example, in US Pat. No. 7,623,909 and US Pat. No. 8,200,341, sensing vectors with similar R and T wave amplitudes are often disadvantageous from the start. Thus, vector selection can be used to select vectors with a larger R: T amplitude ratio.
0113When the two-stage pulse pattern is analyzed using autocorrelation, the outputs either reflect R-wave and T-wave peaks that are alternating and have different morphologies, or have a two-stage pulse pattern. Determining whether it reflects two R waves can be difficult. Vector selection can be used to avoid confusion, along with a set of rules shown and demonstrated in FIG.
0114As shown in 400, the two-stage pulse rhythm produces a pattern of high and low peaks when undergoing the autocorrelation analysis of FIGS. 4-5 above. Using the methods of FIGS. 6-7 above usually results in selecting peak 402 as the RR estimate peak. However, since the actual rhythm is a two-stage pulse signal, the true RR is at half the lag depth of the originally chosen RR estimate, i.e. at peak 404.
0115To address this potential problem, an optional two-stage pulse test is shown below in 408. Not all patients emit this rhythm pattern, so an optional two-stage pulse test can be "switched on" by a physician.
0116The test begins after the RR estimate is calculated at 410. The test then looks for peaks that are spaced half the lag depth of the RR estimate, as shown in 412. Peaks 404 and 406 meet the check at 412, as shown in the graphic at 400.
0117After passing the check at 412, the method is whether the value of the R [n] peak identified in block 412 is within a predetermined proportion of the original value of the R [n] peak selected as the RR estimate. Decide whether or not. "RVal" is used as an abbreviation for the value of R [n] for each peak in the drawing. Here, the threshold is shown in graphic 400, peak 404 exceeds the threshold and passes step 414. The threshold for exemplary purposes is 55% of the RVal peak for the original RR estimate, and other thresholds may be used, for example, in the range of 40-80%.
0118If both 412 and 414 are passed, the two-stage pulse test will paraphrase the RR estimate using the peak at RR estimate / 2, as shown in 416, where the peak 404 is the RR estimate. However, the RVal for the original RR estimate is retained as shown in 416 for the purpose of determining outcome confidence, taking into account the identification of possible two-stage pulse patterns. Therefore, peak 404 has an R [n] value of about 0.5, but the reported RVal value is about 0.75 (R [n] value for peak 402).
0119If either 412 or 414 does not pass, the two-stage pulse test also fails and the method ends at 418. Similarly, the two-stage pulse test ends after any corrections have been made in block 416.
0120Figure 17 addresses a test to check for jitter. Jitter can occur where RR intervals are sometimes inconsistent, leading to split peaks at the output, such as those shown at 430 in the graphic. Using the methods of FIGS. 6-7, RR estimates are identified at 432. However, the split peaks appear at 434, the split peaks occur at about RR estimate / 2, and the R [n] values for each of the split peaks exceed the relative thresholds. It suggests that jitter (or alternans, when changing RR intervals can be required) may be occurring.
0121A method is shown in 436 to test for such jitter. First, the RR estimate is calculated as shown in 438. The method then checks the split peak at the RR estimate / 2, as shown at 440. If such split peaks are found at 440, the method determines whether each of the split peaks meets the Rval threshold, as shown at 442. In the exemplary method, the Rval threshold is 50% and other thresholds can be used, for example, in the range of 40% to 80%.
0122If each of checks 440 and 442 passes, the method would paraphrase the RR estimate as the RR estimate / 2, as shown in 444. Similar to the two-stage pulse test, the Rval from the original RR estimate can be retained in this example. If any of checks 440 and 442 do not pass, the jitter test ends as shown in 446.
0123For each of the two-stage pulse test (Figure 16) and the jitter test (Figure 17), when the test triggers a correction of the RR estimate, a flag can be set, a counter can be incremented, or an event, and so on. If not, it can be counted / identified. In some examples, passing any of these tests can allow the system to store data in memory for later doctors to search and review any such events. is there.
0124Figures 18A-18B show several schemes in which the R [n] computer, peak selector, RR estimate tracker, and treatment decision can be linked together. In the example of FIG. 18A, the R [n] calculator 460 reports the output of the R [n] calculation to the peak selector 462. The peak selector 462 provides the peak tracker 464 with an RR estimate (a) and a set of peaks. The peak selector 462 also provides the treatment decision block 466 with an RR estimate (a), in this example, along with any flags resulting from the peak selector 462 analysis, and a confidence (a) indicator. The treatment decision block 466 uses the RR estimate (a) from the peak selector 462, as well as any flags and reported confidence (a), to determine if the conventional rate estimate is likely to be correct. Or it is possible to determine if it is unlikely to be correct. The RR estimate tracker 464 reports the RR estimate (b) and reliability to treatment decision 466.
0125For example, the peak selector 462 is capable of identifying the RR estimate (a), which has low confidence (a), while the RR estimate tracker 464 is based on a secondary peak. The higher confidence (b) identifies different RR estimates (b), the secondary peaks fill the existing track, and the secondary peaks are the reported RR estimates (a). Either has one or more pickets, or is in the tachycardia zone, even if is not in the track. In that case, treatment decision block 466 can ignore the RR estimate (a) and adopt the RR estimate (b) instead.
0126In another example, the RR estimate (a) is reported with high confidence (a), but the RR estimate tracker does not find a peak in the existing track and reports that it is coasting. However, if the previous stored RR estimate is used and a low confidence (b) is reported, treatment decision block 406 adopts the RR estimate (a) rather than the RR estimate (b). It is possible to do.
0127Therefore, in the example of FIG. 18A, the treatment decision block 466 uses the confidence reports from the peak selector 462 and the RR estimate tracker 464, respectively, with the RR estimates (a) and RR estimates (b). It is permissible to choose between.
0128In the example of FIG. 18B, the R [n] computer 480 again provides the result to the peak selector 482. The peak selector 482 serves its function and provides the peak, RR estimate (a), arbitrary set flag, and reliability (a) to the RR estimate tracker 484. The RR estimate tracker 484 performs its function and provides the RR estimate (b), confidence (b), and any set flag to treatment decision block 486. Therefore, in FIG. 18B, the RR estimate tracker determines a single output RR estimate (a) with associated confidence (b) for treatment decision block 486.
0129One or more of the individual blocks in Figures 18A-18B can be separate pieces of hardware in a single system, but two or more blocks are single. Can be integrated into a dedicated circuit. Alternatively, the separate blocks in FIGS. 18A-18B can be separate functional blocks in a larger software configuration. For example, given a stream of data (or a stored stack), a function call for 460/480 that computes R [n] can be made, followed by a peak selection given the output R [n]. Followed by a function call for 462/482 to perform, followed by a function call for 464/484 to track the RR estimate using the RR estimate (a) from the peak selection and the peak, RR Estimates (a) and peaks can all be used as inputs (with other data) to call treatment decisions 466/486. In one example, blocks 480, 482, and 484 are provided on dedicated circuits, where the output of these blocks is provided to a processor or controller where the treatment decision process is performed.
0130In the embodiments shown in FIGS. 18A-18B (and other examples shown above and below), the RR estimate can be considered a heart rate estimate. If a confidence measure is provided in connection with the RR estimate and one or more peaks, such is as one or more possible estimates of heart rate. Can be treated.
0131FIG. 19 illustrates the integration of the two methods for identifying rates. The conventional rate method is illustrated using the block 500, in which the R waves are individually detected by comparing the detected signal with a threshold. Conventional R-wave detection can be used in block 500. Examples of some exemplary purposes appear in US Pat. No. 8,565,878 and US Pat. No. 5,709,215.
0132The detected R wave is reported to the noise / overdetection removal block 502, which confirms that the R wave is probably a cardiac event. When individually detected R waves are identified in 502, rate and shape (morphology) information is acquired and provided to 504, treatment decision and / or enforcement block 506. This conventional method then returns to standby state 508 until the next R wave detection.
0133The method also integrates rate calculations using autocorrelation, which can be called asynchronously (eg, at fixed intervals) or synchronously to the new detection 500, if desired. This standby state is shown in 510. Upon activation, autocorrelation rate estimates are made using one or more combinations of calculating R [n], selecting peaks, and tracking RR estimates 512. .. The resulting RR estimate is then reported to treatment decision block 506 at 514 and re-entered standby state 510. RR estimates from block 512 can be generated using the tracking tools described above, or in some cases heart rate estimates are directly from the peak selector that evaluates the autocorrelation function. Can be generated Therefore, tracking is explicitly optional in the above example.
0134Treatment decision 506 can use each of these different calculations in different approaches to identify whether treatment is needed. For example, one of the rates can be used to double-check the other, or the rates can be compared to identify a match. If the rates do not match, additional analysis can be performed using additional sensing inputs, such as motion sensors or blood pressure or oxygenation sensors. If the rates both indicate that treatment is needed (whether matching or not), the treatment function can be called. Other approaches are also described above.
0135In one example, if a dominant peak test is applied and filled with a dominant peak, treatment decision 506 will determine the estimated heart rate associated with the dominant peak using R-wave detection from block 500. It may be configured to be treated as more reliable than the generated rate. In another example, peaks that pass the picket test can be treated as more reliable in treatment decision 506 than the rates generated using R-wave detection from block 500. In yet another example, the peak selection output is R only until the track is declared via the method of FIGS. 8A-8B and only if the peak selection output falls within the specified track. It can be treated as less reliable than the wave detection output.
0136Yet another example can have multiple analytical courses, depending on the state of the R wave detection rate tracking and peak selective output. For example, the following rules can be applied in various examples.
0137--If both the R wave detection rate and the autocorrelation rate match and are high, a high rate is confirmed, suggesting tachyarrhythmia. --R wave detection suggests a high rate, but if the autocorrelation rate is low, additional analysis is required before the high rate is treated as valid in any of the following cases: (Wait time, detected event width, or morphology analysis) The autocorrelation rate is based on a rate estimate that falls within a valid track (the rate estimate is either a candidate from the peak analysis or a selected peak), or The autocorrelation rate is based on selected peaks that pass one of the picket test or the dominant peak test. --- If the R-wave detection rate is low, but the autocorrelation rate is high, then the autocorrelation rate is not in the declared track, and (directly or through a large tachycardia peak test). If it is not based on peaks that pass the picket test (regardless of), additional analysis (wait time, detected event width, or morphology analysis) is required before the high rate is treated as valid. In another example, tracking is omitted and the following rules can be applied.
0138--The rates calculated using R-wave detection are high, and if the autocorrelation rate (whether or not the rates match) exceeds the tachycardia threshold. Treated as valid --The rate calculated using R-wave detection is treated as valid if it is high and if the autocorrelation test cannot meet either the picket test or the dominant peak test. To be --The rate calculated using the autocorrelation test is lower than the R wave detection rate and below the tachycardia threshold, and if it passes either the picket test or the dominant peak test. , Treated as valid In addition, other combinations are possible within the scope of the present invention.
0139In one example, if the autocorrelation analysis is called periodically and if the autocorrelation rate is calculated with high confidence, the autocorrelation rate will be the R wave until the next iteration of the autocorrelation analysis. Replaces the rate calculated using detection. In systems that use NID or Y minute X filters, the autocorrelation analysis rate can then be treated as repeating during the period during which the R wave detections are exchanged. For example, if the autocorrelation determines the rate for every 180 beats and the autocorrelation function is called at 1 second intervals, then the NID or Y minute X filter analysis is 1 between the iterative calculations of the autocorrelation analysis. During the second interval, three events will be added at 180 beats per minute.
0140The treatment decision 506 is, for example, several iterative calculations using one rate direct calculation or using one or more of the NID or Y minute X filters as discussed above. It is possible to use a calculation over to determine if the heart rate, as estimated by one or both of blocks 502/512, exceeds the therapeutic threshold. Treatment decisions can be combined with morphology information gathered from cardiac signals and rates. In some examples, treatment decision 506 is a shock-only borderline, where above-threshold rates are considered to require high-energy cardioversion or defibrillation shock, eg, anti-defibrillation. VT zones to which lower energy therapies such as tachycardia pacing are applied, as well as conditional zones where additional analysis of shape elements (template matching, width, spacing stability, amplitude, etc.) and rate combinations is performed. It is possible to set two or more rate boundaries, including one or more of them. The treatment decision 506 is an additional sensor input or an input from a separate device, such as measurements such as blood oxygenation, pressure, color, etc., such as a pressure monitor, reed space maker, etc., from different devices. Measurements, or may be provided separately within the patient, or integrated in a single device with the rest of the system performing the autocorrelation and other functions described above. It is possible to integrate the measured values from the position sensor or the movement sensor.
0141Various notes and examples The first non-limiting example is an implantable medical device system configured for iterative analysis of cardiac signals, with multiple electrodes (16, 18) for sensing cardiac signals. , 20, 36, 38, 40, 42), which is an autocorrelation means for generating an autocorrelation function from the sensed cardiac signal, and the autocorrelation function has an amplitude as a function of lag depth. Correlation means and a peak selector means for identifying amplitude peaks in an autocorrelation function, the first estimate of heart rate, and at least a first with a low or high confidence rating. It takes the form of an implantable medical device system consisting of a peak selector means, which is intended to find relevant reliability. Further, in the first non-limiting embodiment, the peak selector means has at least one additional peak with respect to the selected peak having the first lag depth in the autocorrelation function of a given iterative calculation. It consists of picket test means to determine if the peak appears in a second lag depth, which is a multiple of the first lag depth, and the peak selector means further selects and estimates candidate peaks. It consists of candidate selection means for determining the suitability of candidate peaks for calculating the heart rate to be performed, and the candidate selection means can operate to identify a large number of candidate peaks using the peaks of the autocorrelation function. The candidate selection means is, as the first candidate peak, the following, that is, the candidate peak having the minimum lag depth among the identified candidate peaks, or the candidate peak having the minimum lag depth. Can operate to select one of the candidate peaks, which has an amplitude greater than at least the first margin and corresponds to a heart rate that exceeds the rate threshold, and is a peak selector. The means are configured to use candidate selection means to identify one or more candidate peaks, and picket test means are suitable for any candidate peak to estimate heart rate. Configured to determine if and, if so, report an estimated heart rate It has been. Figure 3 and related texts include autocorrelation means that generate the autocorrelation function R [n] at 60, peak selector means at 62, and blocks 64, 66, and 68, with an estimated rate at 70. The first non-limiting embodiment is illustrated by including and a tracking means that generates reliability. Another example is in FIG. 6, which illustrates a peak selector means, which includes an overall identification or finds a peak in 150 and related text, with an RR interval at 174. Perform an analysis that leads to an estimate of the rate in the form of the estimate. Peak selector means, such as those shown in FIG. 6, select candidates in 154, 156, 158, 160, 162, and related texts and picket testing decisions in 164, 166, 168, and related texts. Including.
0142The second non-limiting example is the implantable medical device system described in the first non-limiting example, in which the peak selector means is one of the candidate peaks, all other. Further including a dominant peak testing means for determining whether the peak is exceeded by at least a second margin, the peak selector means used the dominant peak testing means and was found to be suitable by the picket testing means. In the absence of a candidate peak, it takes the form of an implantable medical device system that can act to attempt to identify a suitable peak for estimating heart rate. Dominant peak testing means are illustrated in 172 in Figure 6 and related text.
0143A third non-limiting example is the implantable medical device system described in the second non-limiting example, in which the dominant peak testing means is at least all other than the second margin. If a dominant peak that exceeds the peak of is identified and the dominant peak corresponds to a heart rate below the dominant peak rate threshold, the peak selector means has an estimated heart rate corresponding to the dominant peak. It takes the form of an implantable medical device system, reported as.
0144A fourth non-limiting example is the implantable medical device system described in the third non-limiting example, in which a cardiac event is generated by comparing an electrical signal from an electrode with a threshold. Treatment is required by taking results from each of the R wave detection means and the R wave detection means and the tracking means for detecting and thereby generating multiple R wave detections and the resulting rate estimates. Further including a deciding means for determining whether or not, the deciding means will make the estimated heart rate reported by the peak selector means based on the dominant peak higher than the rate generated by the R wave detecting means. It takes the form of an implantable medical device system that is configured to be accepted as reliable. The inclusion of R-wave detection and determination means is illustrated in at least FIG. 19, which includes, for example, R-wave detection 500, and determination means in the 506 and related texts.
0145A fifth non-limiting example is the implantable medical device system described in the first or second non-limiting example, in which the candidate selection means is a predetermined in the autocorrelation function. It is configured to select the maximum peak of the quantity, and the peak selector means that the peak in the autocorrelation function at the depth corresponding to the tachycardia rate is the third margin of the maximum peak in the autocorrelation function. It further includes a large tachycardia peak checking means to determine if it is in, and if so, the peak selector means submits the peak identified by the large tachycardia peak checking means to the picket testing means. It takes the form of an implantable medical device system that is configured to determine if it is appropriate for estimating heart rate.
0146The sixth non-limiting example is the implantable medical device system according to any one of the first to third non-limiting examples, which tracks the output of the peak selector means. It takes the form of an implantable medical device system, further including tracking means for generating heart rate estimates from it.
0147The seventh non-limiting example is the implantable medical device system described in the sixth non-limiting example, which identifies any peak in an autocorrelation function greater than the reporting threshold. It takes the form of an implantable medical device system that further includes reporting means for reporting to tracking means.
0148The eighth non-limiting example is the implantable medical device system described in the sixth non-limiting example, which identifies the maximum peak in the autocorrelation function and the threshold of the maximum peak. It takes the form of an implantable medical device system that further includes reporting means for reporting any peak in the autocorrelation function greater than the percentage to the tracking means.
0149The ninth non-limiting example is the implantable medical device system according to any one of the sixth to eighth non-limiting examples, wherein the electric signal from the electrode is used as a threshold value. Results from each of the R-wave detection means and the R-wave detection and tracking means for detecting cardiac events by comparison and thereby generating multiple R-wave detections and the resulting rate estimates. It takes the form of an implantable medical device system, further including a determinant to determine if treatment is needed.
0150The tenth non-limiting example is the implantable medical device system according to any one of the first to third non-limiting examples, wherein the electric signal from the electrode is used as a threshold value. From each of the R-wave detection means and the R-wave detection means and the peak selector means for detecting cardiac events by comparison and thereby generating multiple R-wave detections and the resulting rate estimates. It takes the form of an implantable medical device system that further includes a deciding means for taking results and deciding whether treatment is needed.
0151The eleventh non-limiting example is the implantable medical device system described in the tenth non-limiting example, in which the determinant is the rate estimate reported by the peak selector means. Results from the peak selector means are more reliable than results from the R wave detection means if they are based on candidate peaks identified by the picket test means as appropriate for estimating heart rate. It takes the form of an implantable medical device system that is configured to be treated as.
0152The twelfth non-limiting example is the implantable medical device system according to any one of the first to eleventh non-limiting examples, in which the rate threshold is 75 beats per minute. It takes the form of an implantable medical device system set in.
0153The thirteenth non-limiting example is the implantable medical device system according to any one of the first to twelfth non-limiting examples, in which the autocorrelation means is a series of outputs. Generating an autocorrelation function with sample {1 ... N}, the picket test means whether there are at least two pickets for candidate peaks with a lag depth of less than N / 3, and N / 3. It takes the form of an implantable medical device system configured to identify whether at least one picket is present for a candidate peak with a lag depth greater than or less than N / 2.
0154The fourteenth non-limiting example is the implantable medical device system according to any one of the first to thirteenth non-limiting examples, in which the peak selector means is a tachycardia lag. Determines if there is any peak in the autocorrelation function greater than the tachycardia threshold located in the subthreshold lag depth, and if so, possible tachycardia It takes the form of an implantable medical device system that is configured to set a flag for sexual arrhythmias.
0155A fifteenth non-limiting example is the implantable medical device system according to any one of the first to fourteenth non-limiting examples, at least autocorrelation means and peak selector means. It takes the form of an implantable medical device system consisting of a canister containing an arithmetic circuit including, and a lead system including at least some of a plurality of electrodes.
0156A sixteenth non-limiting example is a method of analyzing a cardiac signal in an implantable medical device, the implantable medical device having a plurality of electrodes for sensing the cardiac signal. The plurality of electrodes are connected to an arithmetic circuit for at least performing analysis of the sensed cardiac signal, and the method is a step for generating an autocorrelation function from the sensed cardiac signal. The autocorrelation function has an amplitude as a function of the lag depth, and the step of identifying the amplitude peak in the autocorrelation function and finding the first estimate of the heart rate, and the lag depth, respectively. At least more than the step of identifying one or more candidate amplitude peaks having and the candidate peak with the smallest lag depth of the identified candidate peaks, or the candidate peak with the smallest lag depth. A first candidate peak with a first lag depth by choosing one of the candidate peaks that has an amplitude larger by the first margin and that corresponds to the heart rate that exceeds the rate threshold. Determines whether at least one additional peak appears in the second lag depth, which is a multiple of the lag depth of the first candidate peak, and if so. By finding that the picket test has passed for the first candidate peak, it takes the form of a method of analyzing a cardiac signal, consisting of a step of applying the picket test to the first candidate peak.
0157The 17th non-limiting example is the method of analyzing the heart signal according to the 16th non-limiting example, in which the first estimate of heart rate is picketed by the first candidate peak. Analyzing the cardiac signals generated by converting the first lag depth to a time interval and by converting the time interval to a rate in response to finding that the test passed. Take the form of a method.
0158The eighteenth non-limiting example is the method of analyzing the cardiac signal according to the sixteenth non-limiting example, and finds that the picket test was not passed for the first candidate peak. It takes the form of a method of analyzing cardiac signals, further comprising a step, a step of selecting a second candidate peak, and a step of applying a picket test to the second candidate peak.
0159The nineteenth non-limiting example is the method of analyzing the cardiac signal according to the sixteenth non-limiting example, in which the picket test was not passed for at least the first candidate peak. The step of finding, the step of identifying the maximum peak in the autocorrelation function, the step of evaluating whether the maximum peak is at least a threshold amount larger than any other peak in the autocorrelation function, and the maximum. A cardiac signal further comprising finding that the peak is at least a threshold amount greater than any other peak and calculating a first estimate of heart rate using the lag depth of the maximum peak. Takes the form of a method of analyzing.
0160The twentieth non-limiting example is the method of analyzing the cardiac signal according to the sixteenth non-limiting example, in which the step of finding that the picket test is passed for the first candidate peak is included. The step of checking whether the autocorrelation function contains a non-candidate peak with a lag depth smaller than that of the first candidate peak and an amplitude within the threshold of the first candidate peak, and so on. In some cases, it takes the form of a method of analyzing cardiac signals, further comprising the step of determining whether the non-candidate peak passes the picket test.
0161The 21st non-limiting example is a method for analyzing a cardiac signal according to any one of the 16th to 20th non-limiting examples, in which R is applied to the sensed cardiac signal. The steps of performing wave detection and generating multiple R-wave detections, using R-wave detection to calculate a second estimate of heart rate, and the first and second estimates of heart rate. It takes the form of a method of analyzing cardiac signals, further comprising the steps of analyzing and determining if treatment is needed.
0162The 22nd non-limiting example is the method of analyzing the heart signal according to any one of the 16th to 21st non-limiting examples, and the first non-limiting example of the heart rate over time. A method of analyzing cardiac signals that further includes the step of tracking estimates and establishing a confidence measure if the first estimate of heart rate is consistent over either a period or a series of calculations. Take the form.
0163The 23rd non-limiting example is the method of analyzing the heart signal according to any one of the 16th to 22nd non-limiting examples, in which the maximum peak in the autocorrelation function is set. Analyze and determine if the picket test is not passed by at least the first candidate peak, determine if it is a dominant peak, the amplitude within the similarity boundary of the first candidate peak, and the first Analyze whether non-candidate peaks that have a smaller lag depth than those of the candidate peaks and are associated with heart rates above the tachycardia threshold appear in the autocorrelation function, and if so The process of generating confidence associated with the first estimate of heart rate, such as determining whether a non-candidate peak passes the picket test, and a picket with at least two candidate or non-candidate peaks. If you pass the picket test with a step that puts a high degree of confidence in the first estimate of heart rate, if a dominant peak is found, or with a picket with only one candidate or non-candidate peak If you pass the picket test, place an intermediate confidence in the first estimate of heart rate, or if neither the dominant peak nor the peak that passes the picket test is found, then in the autocorrelation function The process of generating confidence associated with a first estimate of heart rate, such as reporting an estimated heart rate based on the maximum peak and placing a lower confidence on the first estimate of heart rate. And, further including, peaks that pass the picket test with two peaks have at least the first and second additional peaks at multiple lag depths of the peaks in the analysis in the picket test. It takes the form of a method of analyzing heart signals, which means that.
0164The 24th non-limiting example is the method of analyzing the heart signal according to the 23rd non-limiting example, in which R wave detection is performed on the sensed heart signal and a plurality of heart signals are detected. The process of generating R-wave detection, the process of using R-wave detection to calculate a second estimate of heart rate, and the analysis of the first and second estimates of heart rate, the first estimate. By treating the first estimate of heart rate as more reliable than the second estimate of heart rate, and by treating the first estimate as less reliable, if is associated with high confidence. The process of determining if treatment is needed by treating the second estimate of heart rate as more reliable than the first estimate of heart rate, if accompanied by a degree. It takes the form of a method of analyzing a cardiac signal, further including.
0165A 25th non-limiting example consists of an implantable canister containing an arithmetic circuit for performing cardiac signal analysis and a plurality of electrodes connected to the arithmetic circuit for providing the cardiac signal to the implantable canister. The arithmetic circuit takes the form of an implantable cardiac device configured to perform the method of cardiac signal analysis according to any one of the 16th to 24th non-limiting examples. ..
0166The non-limiting embodiment of any one of the first to the twenty-fifth may further include a two-stage pulse discriminating means or step and / or a jitter discriminating means or step. An example of two-stage pulse identification is shown in Figure 16 and the associated text. In addition, an example of jitter identification is shown in Figure 17 and the associated text.
0167Each of these non-limiting examples may be independent or may be combined with one or more of the other examples in various substitutions or combinations. The detailed description above includes references to the accompanying drawings that form part of the detailed description. The drawings, by way of example, show certain embodiments in which the present invention may be practiced. These embodiments are also referred to herein as "Examples". Such embodiments can include elements in addition to those shown or described. However, we also contemplate examples in which only those elements shown or described are provided. Moreover, we present a particular example (or one or more aspects thereof), or any other example (or one or more thereof) shown or described herein. Examples are also contemplated that use any combination or substitution of those elements shown or described (or one or more aspects thereof) with respect to any of the embodiments).
0168If there is inconsistent usage between this document and any document incorporated herein by reference, the usage of this document will prevail. In this document, the term "a" or "an" shall include one or more, regardless of any other example or usage of "at least one" or "one or more". Is used for. In this document, the term "or" is used to represent a non-exclusive OR, or "A or B" is "A but not B", unless otherwise indicated. , "B but not A", and "A and B" are now included. In this document, "including" and "in" The term "which" is used as a plain English equivalent of the terms "comprising" and "where in" respectively. Also, in the following claims, the terms "including" and "comprising" are open, i.e., in addition to those listed after such terms in the claims, elements. The including system, device, article, composition, formulation, or method is still considered to fall within its claims. Moreover, in the following claims, terms such as "first", "second", and "third" are simply used as labels to impose numerical requirements on their objects. Not intended.
0169The examples of methods described herein can be implemented, at least in part, by a machine or computer. Some examples can include computer-readable or machine-readable media encoded by instructions that can operate to configure an electronic device to perform the method as described in the examples above. Is. Implementations of such methods can include code, such as microcode, assembly language code, or high-level language code. Such code can include computer-readable instructions for implementing various methods. The code can form part of a computer program product. Further, in the example, the code may be tangibly stored on one or more volatile, non-temporary, or non-volatile tangible computer-readable media, for example, during execution or at other times. .. Examples of these tangible computer readable media are hard disks, removable magnetic disks, removable optical disks (eg, compact disks and digital video disks), magnetic cassettes, memory cards or memory sticks, random access memory (RAM), and read-only memory (eg, compact disks and digital video disks). ROM) etc. can be included, but it is not limited to that.
0170The above description is for illustrative purposes only and is intended to be non-limiting. For example, the above examples (or one or more embodiments thereof) may be used in combination with each other. Other embodiments may be used, for example, by one of ordinary skill in the art, by reviewing the above description. The abstract is provided to comply with Section 1.72 (b) of the US Patent Law Enforcement Regulations so that readers can quickly identify the nature of the technical disclosure. The abstract is submitted with the understanding that it will not be used to interpret or limit the scope or meaning of the claims.
0171Also, in the above detailed description, various features can be grouped for efficient disclosure. This should not be construed as intended that the unclaimed disclosed features are mandatory for any claim. Rather, the subject matter of the present invention may be less than all the features of a particular disclosed embodiment. Therefore, the following claims are incorporated as examples or embodiments, thereby in detail, each claim being self-sustaining as a separate embodiment, and such an embodiment. It is contemplated that the forms can be combined with each other in various combinations or substitutions. The scope of the invention, along with the appended claims, should be determined with reference to the full scope of equivalents for which such claims enjoy rights.
22 sheets
Sheet 1 Sheet 2 Sheet 3 Sheet 4 Sheet 5 Sheet 6 Sheet 7 Sheet 8 Sheet 9 Sheet 10 Sheet 11 Sheet 12 Sheet 13 Sheet 14 Sheet 15 Sheet 16 Sheet 17 Sheet 18 Sheet 19 Sheet 20 Sheet 21 Sheet 22
Every citation, both ways
| Document | Relation | Office | Category | Cited during | Relevant claims |
|---|---|---|---|---|---|
| JP2007524466A | Cites | Japan | X | Search report | 1,5-10,13-15 |
| JP2012130391A | Cites | Japan | A | Search report | – |
| US2013079657A1 | Cites | United States of America | A | Search report | – |
| JPH0595914A | Cites | Japan | A | Search report | – |
35 members in 6 offices
Priority claims9
| Document | Office | Kind | Date |
|---|---|---|---|
| 62038440 | United States of America | – | |
| 62038437 | United States of America | – | |
| 62038438 | United States of America | – | |
| 201462038440 | United States of America | P | |
| 201462038437 | United States of America | P | |
| 201462038438 | United States of America | P | |
| 14819851 | United States of America | – | |
| 201514819851 | United States of America | A | |
| 2015045158 | United States of America | W |
Members35
| Document | Office | Kind | |
|---|---|---|---|
| EP2676608A1 | European Patent Office (EPO) | A1 | |
| US2013343521A1 | United States of America | A1 | |
| KR20130142654A | Republic of Korea | A | |
| CN103505228A | China | A | |
| KR101429067B1 | Republic of Korea | B1 | |
| US9125617B2 | United States of America | B2 | |
| US2015342550A1 | United States of America | A1 | |
| US2016045131A1 | United States of America | A1 | |
| US2016045132A1 | United States of America | A1 | |
| US2016045136A1 | United States of America | A1 | |
| WO2016028609A1 | World Intellectual Property Organization (WIPO) | A1 | |
| WO2016028610A2 | World Intellectual Property Organization (WIPO) | A2 | |
| WO2016028611A1 | World Intellectual Property Organization (WIPO) | A1 | |
| WO2016028610A3 | World Intellectual Property Organization (WIPO) | A3 | |
| US9451892B2 | United States of America | B2 | |
| US9451893B2 | United States of America | B2 | |
| US2016367829A1 | United States of America | A1 | |
| US9629565B2 | United States of America | B2 | |
| CN106659409A | China | A | |
| CN106659900A | China | A | |
| EP3182882A2 | European Patent Office (EPO) | A2 | |
| EP3182883A1 | European Patent Office (EPO) | A1 | |
| US2017209063A1 | United States of America | A1 | |
| JP2017525474A | Japan | A | |
| JP2017529902AThis record | Japan | A | |
| US9895071B2 | United States of America | B2 | |
| US9993213B2 | United States of America | B2 | |
| US10016143B2 | United States of America | B2 | |
| EP2676608B1 | European Patent Office (EPO) | B1 | |
| JP6487033B2 | Japan | B2 | |
| CN103505228B | China | B | |
| CN106659409B | China | B | |
| CN106659900B | China | B | |
| EP3182882B1 | European Patent Office (EPO) | B1 | |
| EP3182883B1 | European Patent Office (EPO) | B1 |
12 legal events, as the office reported them to INPADOC
Over the term
Point at a mark for the eventEvents
| Event | Code | |
|---|---|---|
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Certificate of patent or registration of utility modelJAPANESE INTERMEDIATE CODE: R150R150 | R150 | |
| First payment of annual fees (during grant procedure)JAPANESE INTERMEDIATE CODE: A61A61 | A61 | |
| Written decision to grant a patent or to grant a registration (utility model)JAPANESE INTERMEDIATE CODE: A01A01 | A01 | |
| Decision of grant or rejection writtenTRDD | TRDD | |
| Request for written amendment filedJAPANESE INTERMEDIATE CODE: A523A521 | A521 | |
| Written request for extension of timeJAPANESE INTERMEDIATE CODE: A601A601 | A601 | |
| Notification of reasons for refusalJAPANESE INTERMEDIATE CODE: A131A131 | A131 | |
| Written request for application examinationJAPANESE INTERMEDIATE CODE: A621A621 | A621 |
Numbers
- Publication
- 2017529902
- Application
- 2017509761
Titles2
- Japanese
- 植え込み型の医療用デバイスにおける心拍数自己相関分析のためのピーク選択
- English
- Peak selection for heart rate autocorrelation analysis in implantable medical devices
Classification
- CPC, 20
- A61B5/7246
- A61B5/4836
- A61B5/686
- A61N1/3704
- A61N1/3956
- A61N1/3987
- A61B5/0245
- A61N1/3925
- G06F17/16
- A61B5/341
- A61B5/287
- A61B5/35
- A61B5/361
- A61B5/352
- A61B5/363
- A61B5/346
- G16H10/60
- G16H40/67
- G16H20/30
- G06F17/11
- IPC, 9
- A61B5 0452
- A61B5 0456
- A61B5 0408
- A61B5 0478
- A61B5 0492
- A61B5 363
- A61B5 296
- A61B5 352
- A61B5 361
Designated states5
- Regional, 4
- Zimbabwe
- Turkmenistan
- Türkiye
- Togo
- National, 1
- United States of America