Extraction of cardiac signal data

• US 8,433,395 B1
• Filed: 06/29/2011
• Issued: 04/30/2013
• Est. Priority Date: 11/03/2009
• Status: Active Grant

• Current Assignee: VivaQuant LLC
• Original Assignee: VivaQuant LLC
• Inventors: Brockway, Marina, Brockway, Brian
• Primary Examiner(s): GETZOW, SCOTT M
• Application Number:
  US13/172,415
• Time in Patent Office: 671 Days
• Field of Search: 600508-523
• US Class Current: 600/509
• CPC Class Codes: A61B 5/316 Modalities, i.e. specific diagnostic methods
• CPC Class Codes: A61B 5/349 Detecting specific parameters of the electrocardiograph cycle
• CPC Class Codes: A61B 5/366 Detecting abnormal QRS complex, e.g. widening
• CPC Class Codes: A61B 5/7203 for noise prevention, reduction or removal
• CPC Class Codes: A61B 5/7221 Determining signal validity, reliability or quality A61B5/7207 and A61B5/7217 take precedence

Abstract

A T-wave offset point of an ECG signal is provided. In accordance with various example embodiments, a location of a QRS complex in the ECG signal is identified and used to determine a first time window of the ECG signal in which to search for a T-wave offset point. The T-wave offset point is identified within the first time window, and the identified T-wave offset point is provided as an output based upon a noise characteristic of the ECG signal in a second time window that includes at least a portion of the T-wave.

Claims

1. A method for providing a T-wave offset point of an ECG signal, the method comprising:

identifying the location of a QRS complex in the ECG signal;

using the identified location of the QRS complex, determining a first time window of the ECG signal in which to search for the T-wave offset point; and

in a computer circuit,

identifying the T-wave offset point within the first time window, and

providing the identified T-wave offset point as an output based upon a noise characteristic of the ECG signal in a second time window that includes at least a portion of the T-wave of the ECG signal.

2. The method of claim 1, wherein the second time window includes the identified T-wave offset point.

3. The method of claim 1, wherein the second time window includes the identified T-wave offset point, the duration of said second time window is about 30% of the duration of a nominal QT interval of the ECG signal, and the second time window extends beyond the T-wave offset point by about 10% of the nominal QT interval duration.

4. The method of claim 1, wherein the second time window has a duration of about 50 msec and is centered on about the identified T-wave offset point.

5. The method of claim 1, wherein the second time window extends from about the point of the largest deflection of the T-wave from an isoelectric line of the ECG signal to about the T-wave offset point.

6. The method of claim 1, wherein the second time window includes a time period extending from about a QRS offset point in the ECG signal to about the T-wave offset point.

7. The method of claim 1, wherein the second time window includes a time period extending the full duration of a cardiac cycle in the ECG signal.

8. The method of claim 1, further comprising computing the noise characteristic by:

decomposing the ECG signal into subcomponents,

identifying said subcomponents in the second time window as either primarily associated with noise or primarily associated with the T-wave of the underlying ECG signal,

computing a noise signal for the second time window by using at least one subcomponent not primarily associated with the T-wave, and

computing said noise characteristic based upon at least one of;

energy of the noise signal, standard deviation of the noise signal, zero crossing density of the noise signal, a metric of noise amplitude computed using an envelope of the noise signal, and a metric of energy of the ECG signal relative to energy of the noise signal.

9. The method of claim 8, further including computing a denoised signal by combining at least two of the subcomponents that are primarily associated with the T-wave, wherein the metric of energy of the ECG signal relative to energy of the noise signal is a signal-to-noise ratio of the ECG signal within the second time window computed as the ratio of respective energies of the denoised and noise signals.

10. The method of claim 1, further comprising computing the noise characteristic by:

computing a denoised signal for the second time window using at least one of a band-pass filter, wavelet thresholding, and an adaptive filter that passes primarily T-wave energy,

capturing a residual of the step of computing the denoised signal as a noise signal for the second time window, and

computing said noise characteristic based upon at least one of;

energy of the noise signal, standard deviation of the noise signal, zero crossing density of the noise signal, a metric of noise amplitude computed using an envelope of the noise signal, and a metric of energy of the denoised signal relative to energy of the noise signal.

11. The method of claim 1, wherein providing the identified T-wave offset point as an output based upon a noise characteristic of the ECG signal in the second time window includes comparing the noise characteristic to a threshold and providing the identified T-wave offset point based upon the comparison.

12. The method of claim 1, wherein providing the identified T-wave offset point as an output includes providing the identified T-wave offset point based upon a presence, in a cardiac cycle within the ECG, of at least one of:

atrial fibrillation, QT dispersion in a multi-lead ECG exceeding a threshold, T-wave morphology complexity exceeding a predefined threshold, ventricular ectopy, a QT interval measurement that falls outside of a user-defined physiologic outlier value or a statistical outlier value.

13. The method of claim 1, further including using the provided identified T-wave offset point to assemble a time series of provided QT interval values for analysis of variability.

14. A method for providing a repolarization signal for a cardiac cycle of an ECG signal, the method comprising:

identifying the location of a QRS complex in the cardiac cycle;

using the identified location of the QRS complex, identifying T-wave onset and offset points;

using the T-wave onset and offset points to respectively define the start and end of the repolarization signal for the cycle; and

in a computer circuit,

determining a noise characteristic of the ECG signal in a time window spanning from about the start to about the end of the repolarization signal, and

providing the repolarization signal as an output, based upon the determined noise characteristic.

15. The method of claim 14, wherein determining the noise characteristic includes:

decomposing the ECG signal into subcomponents,

identifying ones of said subcomponents of the ECG signal within the time window as primarily associated with the T-wave of the underlying ECG signal,

computing a noise signal for the time window using at least one of the subcomponents not primarily associated with the T-wave, and

computing said noise characteristic based upon at least one of;

energy of the noise signal, standard deviation of the noise signal, zero crossing density of the noise signal, a metric of noise amplitude computed using an envelope of the noise signal, and a metric of energy of the ECG signal relative to energy of the noise signal.

16. The method of claim 15, further including computing a denoised signal by combining at least two of the subcomponents that are primarily associated with the T-wave, wherein the metric of energy of the ECG signal relative to energy of the noise signal is a signal-to-noise ratio computed as the ratio of respective energies of the denoised and noise signals.

17. The method of claim 14, wherein determining a noise characteristic includes computing the noise characteristic by:

computing a denoised signal for the time window using at least one of a band-pass filter, wavelet thresholding, and an adaptive filter that passes primarily T-wave energy,

computing a noise signal for the time window using a residual of the step of computing the denoised signal, and

computing said noise characteristic based upon at least one of;

energy of the noise signal, standard deviation of the noise signal, zero crossing density of the noise signal, a metric of noise amplitude computed using an envelope of the noise signal, and a metric of energy of the denoised signal relative to energy of the noise signal.

18. The method of claim 14, wherein providing the repolarization signal as an output, based upon the determined noise characteristic includes comparing the noise characteristic to a threshold and providing the repolarization signal as an output based upon the comparison.

19. The method of claim 14, wherein providing a repolarization signal as an output includes providing the repolarization signal as an output based upon the determined noise characteristic and the presence in the cardiac cycle of at least one of:

atrial fibrillation in the ECG signal, a degree of QT dispersion exceeding a threshold when the ECG signal is a multi-lead signal, T-wave morphology complexity of the ECG signal exceeding a threshold, ventricular ectopy, and a QT interval measurement that falls outside of a user-defined physiologic outlier value or a statistical outlier value.

20. The method of claim 14, further including appending the provided repolarization signal to a matrix of repolarization signals in which a dimension of the matrix corresponds to the number of cardiac cycles of the ECG signal.

21. A method for providing a time series of beat-to-beat QT interval values from a digitized ECG signal of an ambulatory subject, the method comprising:

identifying the location of a QRS complex and a Q-wave onset point of a cardiac cycle of the ECG signal;

using the identified location of the QRS complex, determining a first time window of the cardiac cycle in which to search for a T-wave offset point for a T-wave in the cardiac cycle;

identifying the T-wave offset point within the first time window; and

in a computer circuit,

computing a QT interval value using the identified Q-wave onset point of the QRS complex and the identified T-wave offset point, and

including the QT interval value in a time series of beat-to-beat QT interval values, based upon a noise characteristic of the ECG signal in a second time window that includes at least a portion of the T-wave.

22. The method of claim 21,

further including suppressing energy that is not primarily associated with T-wave energy in the first time window of the digitized ECG signal, prior to identifying the T-wave offset point, using at least one of MDSP denoising, wavelet threshold denoising, band-pass filtering, and adaptive filtering.

23. The method of claim 21, further including computing the noise characteristic by:

decomposing the ECG signal into subcomponents,

identifying said subcomponents as primarily associated with either noise or a T-wave of an underlying ECG signal,

computing a noise signal by combining the subcomponents within the second time window that are primarily associated with noise,

computing a denoised signal by combining the subcomponents within the second time window that are primarily associated with the T-wave of the underlying ECG signal, and

computing said noise characteristic based upon at least one of;

energy of the noise signal, standard deviation of the noise signal, zero crossing density of the noise signal, a metric of noise amplitude based upon an envelope of the noise signal, and a metric of energy of the denoised signal relative to energy of the noise signal.

24. The method of claim 23, wherein the metric of energy of the denoised signal relative to energy of the noise signal is a signal-to-noise ratio.

25. The method of claim 21, further including computing the noise characteristic by:

computing a denoised signal for the second time window using at least one of a band-pass filter, wavelet thresholding, and an adaptive filter, and computing a noise signal as a difference between the digitized ECG and the denoised signal, and

computing said noise characteristic based upon at least one of;

energy of the noise signal, standard deviation of the noise signal, zero crossing density of the noise signal, a metric of noise amplitude computed using an envelope of the noise signal, and a metric of energy of the denoised signal relative to energy of the noise signal.

26. The method of claim 21, wherein including the QT interval value in the time series of beat-to-beat QT interval values includes including the QT interval value based upon a comparison of the noise characteristic to a threshold value.

27. The method of claim 21, wherein including the QT interval value in the time series of beat-to-beat QT interval values includes including the QT interval based upon the noise characteristic and the presence, in the cardiac cycle, of at least one of:

atrial fibrillation, ventricular ectopy, QT dispersion in a multi-lead ECG exceeding a threshold, T-wave morphology complexity exceeding a predefined threshold, ventricular ectopy, a QT interval measurement that falls outside of a user-defined physiologic outlier value or a statistical outlier value.

28. The method of claim 21,

wherein the steps are carried out in the computer circuit executing instructions to carry out the various functions,

further including, in the computer circuit, executing instructions to repeat the steps to compute a plurality of the QT interval values, and

wherein including the QT interval value in a time series of beat-to-beat QT interval values includes, in the computer circuit, including ones of the QT interval values in the time series of beat-to-beat QT interval values based upon a comparison of a noise characteristic of a corresponding ECG signal in the second window to a predefined threshold.

29. The method of claim 21, wherein the time series of beat-to-beat QT intervals consists of QT interval values having error due to noise that is less than 2.5% of the mean QT interval of the ECG signal.