Electrocardiogram noise reduction using multi-dimensional filtering

• US 5,827,195 A
• Filed: 05/09/1997
• Issued: 10/27/1998
• Est. Priority Date: 05/09/1997
• Status: Expired due to Term

• Current Assignee: Spacelabs Healthcare, Inc. (OSI Systems Incorporated)
• Original Assignee: Cambridge Heart Inc.
• Inventors: Lander, Paul
• Primary Examiner(s): Kamm, William E.
• Assistant Examiner(s): LAYNO, CARL HERNANDZ
• Application Number:
  US08/853,914
• Time in Patent Office: 536 Days
• Field of Search: 600/509, 600/516, 600/517
• US Class Current: 600/509
• CPC Class Codes: A61B 5/349 Detecting specific parameters of the electrocardiograph cycle
• CPC Class Codes: A61B 5/36 Detecting PQ interval, PR interval or QT interval
• CPC Class Codes: A61B 5/7203 for noise prevention, reduction or removal
• CPC Class Codes: A61B 5/726 using Wavelet transforms

Abstract

Noise is reduced from a received ECG signal representative of activity of the heart of a patient. A collection of beats from the ECG signal is selected and transformed into a multi-dimensional representation. A multi-dimensional filter function is applied to the multi-dimensional representation to enhance a signal-to-noise ratio of the collection of beats.

Claims

1. A method for reducing noise from an ECG signal representative of activity of the heart of a patient, the method comprising:

altering a physiologic condition of a patient using non-surgical techniques to stress the heart of the patient;

receiving an ECG signal representative of activity of the heart of the patient whose physiologic condition is altered, the ECG signal including a sequence of beats;

selecting a collection of beats from the ECG signal;

transforming the collection of beats into a multi-dimensional representation; and

applying a multi-dimensional filter function to the multi-dimensional representation to enhance a signal-to-noise ratio of the collection of beats.

2. The method of claim 1, wherein:

the step of altering the physiologic condition of the patient comprises instructing the patient to engage in normal physical activity for an extended period, and

the step of receiving an ECG signal comprises receiving an ECG signal recorded using an ambulatory monitor.

3. The method of claim 1, wherein the step of altering the physiologic condition of the patient comprises subjecting the patient to physiologic stress testing.

4. The method of claim 1, wherein the step of selecting a collection of beats comprises selecting a collection of beats that occupy a particular time epoch.

5. The method of claim 1, wherein the step of selecting a collection of beats comprises selecting a collection of beats characterized by ECG shape, wave components, or wave timing.

6. The method of claim 1, wherein the step of selecting a collection of beats comprises selecting a collection of beats having a particular contextual relationship to irregular beats.

7. The method of claim 1, wherein the step of selecting a collection of beats comprises selecting a collection of beats based on patterns of beat-to-beat variation.

8. The method of claim 7, wherein the step of selecting the collection of beats comprises selecting the collection of beats based on patterns of beat-to-beat variation including alternating components.

9. The method of claim 7, wherein the step of selecting the collection of beats comprises selecting the collection of beats based on patterns of beat-to-beat variation including N-geminy components.

10. The method of claim 1, wherein the transformation step includes application of a wavelet transform to obtain a multi-dimensional, time-scale representation of the collection of beats.

11. The method of claim 1, wherein the transformation step includes using a numerical basis set derived from an electrocardiogram signal to obtain a multi-dimensional representation.

12. The method of claim 1, wherein the transformation step includes use of a Wigner-Ville distribution to obtain quadratic multi-dimensional representations of the collection of ECG beats.

13. The method of claim 1, further comprising enhancing the multidimensional representation to reduce variance and interference terms by means of convolution with a multi-dimensional function.

14. The method of claim 13, wherein enhancing the multi-dimensional representation includes convolution with a two-dimensional Gaussian function.

15. The method of claim 13, wherein enhancing the multi-dimensional representation includes convolution with a time-frequency kernel derived from an electrocardiogram signal.

16. The method of claim 13, wherein enhancing the multi-dimensional representation includes convolution with an adaptive, signal-dependent kernel.

17. The method of claim 1, wherein transforming the collection of beats into the multi-dimensional representation comprises transforming the collection of beats into a two-dimensional representation.

18. The method of claim 17, wherein applying the multi-dimensional filter comprises applying a two-dimensional filter.

19. The method of claim 1, further comprising generating the multidimensional filter function.

20. The method of claim 19, wherein generating the filter includes incorporation of any beat from the collection of beats being processed in an a posteriori computation of a filtered signal estimate.

21. The method of claim 19, wherein generating the filter includes incorporation of a priori information about the ECG signal.

22. The method of claim 21, wherein the a priori information is derived from beats which are not part of the collection being processed.

23. The method of claim 21, wherein the a priori information is derived from sources other than the ECG recording of the patient being processed.

24. The method of claim 1, wherein the application step includes performing an inverse transformation of the multi-dimensional representation after filtering to obtain a one-dimensional signal estimate.

25. The method of claim 24, wherein performing the inverse transformation comprises performing an inverse wavelet transform.

26. The method of claim 24, wherein performing the inverse transformation comprises performing an inverse Wigner-Ville transformation.

27. The method of claim 1, further comprising generating a measure of alternans from the enhanced collection of beats.

28. The method of claim 1, further comprising generating a localized ECG signal using the enhanced collection of beats.

29. The method of claim 1, further comprising generating a measure of myocardial ischemia using the enhanced collection of beats.

30. The method of claim 1, further comprising generating a localized ECG signal from the received ECG signal, wherein the step of selecting the collection of beats comprises selecting the collection of beats from the localized ECG signal.

31. The method of claim 1, further comprising calculating an average of the collection of beats, wherein the step of transforming comprises transforming the average of the collection of beats into a two-dimensional representation.

32. A method for measuring alternans in an ECG signal, the method comprising:

receiving an ECG signal representative of activity of a heart of a patient, the ECG signal including a sequence of beats;

selecting a collection of beats from the ECG signal;

transforming the collection of beats into a multi-dimensional representation;

applying a multi-dimensional filter function to the multi-dimensional representation to enhance a signal-to-noise ratio of the collection of beats; and

generating a measure of alternans from the enhanced collection of beats.

33. A method for generating a localized ECG signal, the method comprising:

receiving an ECG signal representative of activity of a heart of a patient, the ECG signal including a sequence of beats;

selecting a collection of beats from the ECG signal;

transforming the collection of beats into a multi-dimensional representation;

applying a multi-dimensional filter function to the multi-dimensional representation to enhance a signal-to-noise ratio of the collection of beats; and

generating a localized ECG signal from the enhanced collection of beats.

34. A method for generating a collection of localized beats, the method comprising:

receiving an ECG signal representative of activity of a heart of a patient, the ECG signal including a sequence of beats;

generating a localized ECG signal from the received ECG signal;

selecting a collection of localized beats from the localized ECG signal;

transforming the collection of localized beats into a multi-dimensional representation; and

applying a multi-dimensional filter function to the multi-dimensional representation to enhance a signal-to-noise ratio of the collection of localized beats.

35. A method for measuring an interval in an ECG signal, the method comprising:

receiving an ECG signal representative of activity of a heart of a patient, the ECG signal including a sequence of beats;

selecting a collection of beats from the ECG signal;

transforming the collection of beats into a multi-dimensional representation;

applying a multi-dimensional filter function to the multi-dimensional representation to enhance a signal-to-noise ratio of the collection of beats; and

generating a measure of the interval from the enhanced collection of beats.

36. The method of claim 35, wherein generating the measure of the interval comprises generating the measure of a QT interval.

37. The method of claim 36, wherein the ECG signal is received from a first lead, the method further comprising:

repeating the receiving, selecting, transforming, applying and generating steps for ECG signals from additional leads; and

comparing the generated measures of intervals for the first and additional leads to determine a measure of QT dispersion.