Method and apparatus for discriminating P and R waves

• US 5,778,881 A
• Filed: 12/04/1996
• Issued: 07/14/1998
• Est. Priority Date: 12/04/1996
• Status: Expired due to Term

• Current Assignee: Medtronic Incorporated (Medtronic Plc)
• Original Assignee: Medtronic Incorporated (Medtronic Plc)
• Inventors: Combs, William J., Sun, Weimin, Panken, Eric J.
• Primary Examiner(s): Kamm, William E.
• Application Number:
  US08/759,494
• Time in Patent Office: 587 Days
• Field of Search: 128/696, 128/920, 128/702
• US Class Current: 600/509
• CPC Class Codes: A61B 5/287 Holders for multiple electrodes, e.g. electrode catheters for electrophysiological study [EPS]
• CPC Class Codes: A61B 5/363 Detecting tachycardia or bradycardia
• CPC Class Codes: A61B 5/7203 for noise prevention, reduction or removal
• CPC Class Codes: A61B 5/7225 Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation A61B5/30, A61B5/316 take precedence; transformers for specific applications H01F38/00; amplifiers H03F
• CPC Class Codes: A61B 5/726 using Wavelet transforms
• CPC Class Codes: A61B 5/7267 involving training the classification device
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients

Abstract

This is a method and apparatus for the automated discrimination of cardiac events of interest, including P-waves, R-waves, T-waves, and specific arrhythmic sequences, in EGM signals for data storage in an implantable monitor or to control operations of an implantable cardiac stimulator through the use of Hidden Markov Modeling techniques and a reduced set of observations. The number of computations and computation time during a heart cycle is reduced by timing the frames of A-EGM samples to the detection of A-SENSE events by the atrial sense amplifier. The A-EGM sample frame is defined in a window preceding and following each A-SENSE event. The A-EGM sample frames are wavelet transformed, and a number of selected W.T. coefficients for each sample frame are saved in a buffer. Each set of saved W.T. coefficients therefore represents either a P-wave or an intrinsic or paced far field R-wave (including fusion beats) unless noise continuously causes A-SENSE events to occur. When the V-SENSE event occurs, an R-trigger is generated, and each set of saved W.T. coefficients is subjected to the HMM algorithm for a determination as to whether the preceding (and any concurrent) A-SENSE events from which the saved W.T. coefficients were derived are P-waves or R-waves. In addition, when the V-SENSE occurs, the HMM algorithm determines stochastically whether or not the successive sets of saved W.T. coefficients represent P-P sequences, far field R-R sequences or P-R sequences or the like.

Claims

1. A method of identifying P-waves and R-waves in an electrocardiogram using Hidden Markov Modeling (HMM), wherein P-waves and R-waves may each be characterized as a state separated by state transitions in a hidden state sequence and wherein there are discrete probabilities that the states will transition from one to the other in a predetermined order, the method comprising the steps of:

sensing the electrocardiogram from at least one electrode;

continuously sampling the sensed electrocardiogram at a predetermined sampling rate and providing a sample value at each sample time;

detecting an event of interest comprising one of the P-wave or R-wave in the electrocardiogram;

framing a sample data set of sample values as a data frame in response to the detection of an event of interest;

wavelet transforming the data frame of sample values to generate m wavelet transformed coefficients;

selecting a sub-set of wavelet coefficients from among the wavelet coefficients representing an observation vector correlated to each data frame;

applying the HMM algorithm to each observation vector to generate the hidden state sequence; and

from the hidden state sequence, determining whether the event of interest is a P-wave or an R-wave.

2. The method of claim 1, wherein the event detection step further comprises:

defining a detection criteria threshold amplitude ensuring that both P-wave and R-wave sample value amplitudes exceed the detection threshold amplitude, and

comparing the sample value amplitudes to the detection criteria threshold amplitude and providing an event trigger signifying the occurrence of an event of interest comprising one of the P-wave or R-wave in the intracardiac electrogram when the sample value amplitude exceeds the detection criteria threshold amplitude.

3. The method of claim 2, wherein the framing step further comprises:

retaining a sample data set of at least m sample value amplitudes on a FIFO basis, and

in response to detection of an event of interest, framing a sample data set of m sample value amplitudes preceding and following the sample value amplitude satisfying the detection criteria as a data frame.

4. The method of claim 3, further comprising the steps of:

determining the cardiac cycle from a feature of the electrogram;

retaining the observation vectors correlated to each data frame through each cardiac cycle;

applying the HMM algorithm to each observation vector retained through the cardiac cycle to generate the hidden state sequence; and

from the hidden state sequence, determining the sequence of P-waves and R-waves in the preceding heart cycle.

5. A method of identifying P-waves and far field R-waves in an atrial electrogram using Hidden Markov Modeling (HMM), wherein P-waves and far-field R-waves may each be characterized as a state separated by state transitions in a hidden state sequence and wherein there are discrete probabilities that the states will transition from one to the other in a predetermined order in a cardiac cycle, the method comprising the steps of:

sensing the atrial electrocardiogram from at least one electrode positioned in sensing relation to the patient'"'"'s atria;

continuously sampling the sensed atrial electrogram at a predetermined sampling rate and providing an atrial sense sample value at each sample time;

detecting an event of interest comprising one of the P-wave or far field R-wave in the atrial electrogram;

framing a sample data set of m atrial sense sample values as a data frame in response to the detection of an event of interest;

wavelet transforming the data frame of m atrial sense sample values to generate m wavelet transformed coefficients;

selecting C_W wavelet coefficients from among the m wavelet coefficients representing an observation vector correlated to each data frame;

retaining the observation vectors correlated to each data frame through a cardiac cycle;

determining the end of a cardiac cycle correlated to the far field R-wave;

upon determining the end of the cardiac cycle, applying the HMM algorithm to each observation vector retained until the end of the cardiac cycle to generate the hidden state sequence; and

from the hidden state sequence, determining whether the event of interest is a P-wave or an R-wave.

6. The method of claim 5, wherein the event detection step further comprises:

defining a P/R-Trigger threshold amplitude ensuring that both P-wave and R-wave representative atrial sense sample value amplitudes exceed the P/R-Trigger threshold amplitude;

comparing the atrial sense sample value amplitudes to the P/R-Trigger threshold amplitude; and

providing a P/R-Trigger signifying the occurrence of an event of interest comprising one of the P-wave or far field R-wave in the atrial electrogram when the atrial sense sample value amplitude exceeds the P/R-Trigger threshold amplitude.

7. The method of claim 6, wherein the framing step further comprises:

retaining a sample data set of at least m atrial sense sample value amplitudes on a FIFO basis; and

in response to a P/R-Trigger, framing a sample data set of m sample value amplitudes preceding and following the atrial sense sample value amplitude satisfying the P/R-Trigger threshold amplitude as a data frame.

8. The method of claim 7, wherein the step of determining the end of the cardiac cycle further comprises:

sensing the ventricular electrogram from at least one electrode positioned in sensing relation to the patient'"'"'s ventricles;

detecting the R-wave from the sensed ventricular electrogram; and

providing an R-trigger from the detected R-wave signifying the end of the cardiac cycle.

9. The method of claim 7, wherein the step of determining the end of the cardiac cycle further comprises:

sensing the ventricular electrogram from at least one electrode positioned in sensing relation to the patient'"'"'s ventricles;

continuously sampling the sensed ventricle electrogram at a predetermined sampling rate and providing a ventricular sense sample amplitude at each sample time;

defining an R-Trigger threshold amplitude ensuring that the R-wave peak sample amplitude exceeds the R-Trigger threshold amplitude; and

comparing the amplitude of each ventricular sense sample amplitude to the R-Trigger threshold amplitude and providing an R-Trigger signifying the end of the cardiac cycle when the ventricular sense sample amplitude exceeds the R-Trigger threshold amplitude.

10. An apparatus for identifying P-waves and R-waves in an electrocardiogram using Hidden Markov Modeling (HMM), wherein P-waves and R-waves may each be characterized as a state separated by state transitions in a hidden state sequence and wherein there are discrete probabilities that the states will transition from one to the other in a predetermined order, the apparatus further comprising:

means for sensing the electrocardiogram from at least one electrode;

means for continuously sampling the sensed electrocardiogram at a predetermined sampling rate and providing a sample value at each sample time;

means for detecting an event of interest comprising one of the P-wave or R-wave in the electrocardiogram;

means for means for framing a sample data set of sample values as a data frame in response to the detection of an event of interest;

means for wavelet transforming the data frame of sample values to generate m wavelet transformed coefficients;

means for selecting a sub-set of wavelet coefficients from among the wavelet coefficients representing an observation vector correlated to each data frame;

means for applying the HMM algorithm to each observation vector to generate the hidden state sequence; and

from the hidden state sequence, determining whether the event of interest is a P-wave or an R-wave.

11. The apparatus of claim 10, wherein the event detection means further comprises:

means for defining a detection criteria threshold amplitude ensuring that both P-wave and R-wave sample value amplitudes exceed the detection threshold amplitude; and

means for comparing the sample value amplitudes to the detection criteria threshold amplitude and providing an event trigger signifying the occurrence of an event of interest comprising one of the P-wave or R-wave in the intracardiac electrogram when the sample value amplitude exceeds the detection criteria threshold amplitude.

12. The apparatus of claim 11, wherein the framing means further comprises:

means for retaining a sample data set of at least m sample value amplitudes on a FIFO basis; and

means responsive to detection of an event of interest for framing a sample data set of m sample value amplitudes preceding and following the sample value amplitude satisfying the detection criteria as a data frame.

13. The apparatus of claim 12, further comprising:

means for determining the cardiac cycle from a feature of the electrogram;

means for retaining the observation vectors correlated to each data frame through each cardiac cycle;

means for applying the HMM algorithm to each observation vector retained through the cardiac cycle to generate the hidden state sequence; and

means for determining the sequence of P-waves and R-waves in the preceding heart cycle from the hidden state sequence.

14. An apparatus for identifying P-waves and far field R-waves in an atrial electrogram using Hidden Markov Modeling (HMM), wherein P-waves and R-waves may each be characterized as a state separated by state transitions in a hidden state sequence and wherein there are discrete probabilities that the states will transition from one to the other in a predetermined order in a cardiac cycle, the apparatus comprising:

means for sensing the atrial electrocardiogram from at least one electrode positioned in sensing relation to the patient'"'"'s atria;

means for continuously sampling the sensed atrial electrogram at a predetermined sampling rate and providing an atrial sense sample value at each sample time;

means for detecting an event of interest comprising one of the P-wave or far field R-wave in the atrial electrogram;

means for framing a sample data set of m atrial sense sample values as a data frame in response to the detection of an event of interest;

means for wavelet transforming the data frame of m atrial sense sample values to generate m wavelet transformed coefficients;

means for selecting C_W wavelet coefficients from among the m wavelet coefficients representing an observation vector correlated to each data frame;

means for retaining the observation vectors correlated to each data frame through a cardiac cycle;

means for determining the end of a cardiac cycle correlated to the far field R-wave;

means operable upon determining the end of the cardiac cycle for applying the HMM algorithm to each observation vector retained until the end of the cardiac cycle to generate the hidden state sequence; and

means for determining from the hidden state sequence whether the event of interest is a P-wave or an R-wave.

15. The apparatus of claim 14, wherein the event detection means further comprises:

means for defining a P/R-Trigger threshold amplitude ensuring that both P-wave and R-wave representative atrial sense sample value amplitudes exceed the P/R-Trigger threshold amplitude;

means for comparing the atrial sense sample value amplitudes to the P/R-Trigger threshold amplitude; and

means for providing a P/R-Trigger signifying the occurrence of an event of interest comprising one of the P-wave or far field R-wave in the atrial electrogram when the atrial sense sample value amplitude exceeds the P/R-Trigger threshold amplitude.

16. The apparatus of claim 15, wherein the framing means further comprises:

means for retaining a sample data set of at least m atrial sense sample value amplitudes on a FIFO basis; and

means responsive to a P/R-Trigger for framing a sample data set of m sample value amplitudes preceding and following the atrial sense sample value amplitude satisfying the P/R-Trigger threshold amplitude as a data frame.

17. The apparatus of claim 16, wherein the means for determining the end of the cardiac cycle further comprises:

means for sensing the ventricular electrogram from at least one electrode positioned in sensing relation to the patient'"'"'s ventricles;

means for detecting the R-wave from the sensed ventricular electrogram; and

means for providing an R-trigger from the detected R-wave signifying the end of the cardiac cycle.

18. The apparatus of claim 16, wherein the means for determining the end of the cardiac cycle further comprises:

means for sensing the ventricular electrogram from at least one electrode positioned in sensing relation to the patient'"'"'s ventricles;

means for continuously sampling the sensed ventricle electrogram at a predetermined sampling rate and providing a ventricular sense sample amplitude at each sample time;

means for defining an R-Trigger threshold amplitude ensuring that the R-wave peak sample amplitude exceeds the R-Trigger threshold amplitude; and

means for comparing the amplitude of each ventricular sense sample amplitude to the R-Trigger threshold amplitude and providing an R-Trigger signifying the end of the cardiac cycle when the ventricular sense sample amplitude exceeds the R-Trigger threshold amplitude.