Electrocardiogram signal detection

• US 9,254,095 B2
• Filed: 11/08/2013
• Issued: 02/09/2016
• Est. Priority Date: 11/08/2012
• Status: Active Grant

• Current Assignee: AliveCor Inc.
• Original Assignee: AliveCor Inc.
• Inventors: Galloway, Conner Daniel Cross, Albert, David E.
• Primary Examiner(s): Flory, Christopher A
• Application Number:
  US14/076,076
• Publication Number:
  US 20140128758A1
• Time in Patent Office: 823 Days
• Field of Search: 600/516, 600/517, 600/518, 600/521
• US Class Current: 1/1
• CPC Class Codes: A61B 5/0006 ECG or EEG signals
• CPC Class Codes: A61B 5/02438 with portable devices, e.g. worn by the patient
• CPC Class Codes: A61B 5/316 Modalities, i.e. specific diagnostic methods
• CPC Class Codes: A61B 5/332 Portable devices specially adapted therefor
• CPC Class Codes: A61B 5/339 Displays specially adapted therefor
• CPC Class Codes: A61B 5/352 Detecting R peaks, e.g. for synchronising diagnostic apparatus; Estimating R-R interval
• CPC Class Codes: A61B 5/361 Detecting fibrillation
• CPC Class Codes: A61B 5/363 Detecting tachycardia or bradycardia
• CPC Class Codes: A61B 5/364 Detecting abnormal ECG interval, e.g. extrasystoles, ectopic heartbeats
• 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

Abstract

Apparatuses and methods for extracting, de-noising, and analyzing electrocardiogram signals. Any of the apparatuses described herein may be implemented as a (or as part of a) computerized system. For example, described herein are apparatuses and methods of using them or performing the methods, for extracting and/or de-noising ECG signals from a starting signal. Also described herein are apparatuses and methods for analyzing an ECG signal, for example, to generate one or more indicators or markers of cardiac fitness, including in particular indicators of atrial fibrillation. Described herein are apparatuses and method for determining if a patient is experiencing a cardiac event, such as an arrhythmia.

Claims

1. A non-transitory computer-readable storage medium storing a set of instructions capable of being executed by a processor, that when executed by the processor, causes the processor to extract an electrocardiogram (ECG) signal from a signal containing noise by causing the processor to:

identify putative QRS regions in the signal containing noise;

cross-correlate the putative QRS regions to determine correlated QRS regions in the signal containing noise;

de-noise the correlated QRS regions by filtering correlated QRS regions in the signal containing noise using a first filtering regime, wherein the correlated QRS regions are filtered by the first filtering regime in isolation of the entire signal containing noise, and filtering a remainder of the signal containing noise outside of the correlated QRS regions using a second filtering regime that is different from the first filtering regime;

construct a de-noised ECG signal comprising the filtered correlated QRS regions and the filtered remainder of the signal outside of the correlated QRS regions; and

present the de-noised ECG signal.

2. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions, when executed by the processor, further causes the processor to:

identify putative R-R intervals in the de-noised ECG signal;

cross-correlate the putative R-R intervals to determine correlated R-R intervals in the de-noised ECG signal; and

modify the de-noised ECG signal by filtering only the correlated R-R intervals of the de-noised ECG signal with a third filtering regime.

3. The non-transitory computer-readable storage medium of claim 2, wherein the set of instructions causes the processor to identify putative R-R intervals in the de-noised ECG signal by constructing an R-R matrix of normalized R-R intervals from the de-noised ECG signal.

4. The non-transitory computer-readable storage medium of claim 3, wherein the set of instructions causes the processor to cross-correlate the putative R-R intervals to determine correlated R-R intervals in the de-noised ECG signal by cross-correlating every normalized R-R interval in the R-R matrix against every other normalized R-R interval in the matrix.

5. The non-transitory computer-readable storage medium of claim 4, wherein the set of instructions causes the processor to determine correlated R-R intervals when a correlation coefficient threshold between R-R intervals is above an R-R correlation threshold.

6. The non-transitory computer-readable storage medium of claim 5, wherein the R-R correlation coefficient threshold is about 0.7.

7. The non-transitory computer readable storage medium of claim 5, wherein the set of instructions causes the processor to modify the de-noised ECG signal by filtering only the correlated R-R intervals of the de-noised ECG signal with a Principle Component Analysis (PCA) of the correlated R-R intervals.

8. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions, when executed by the processor, further causes the processor to pre-filter the signal containing noise before identifying putative QRS regions in the signal containing noise.

9. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions, when executed by the processor, further causes the processor to remove 50 Hz or 60 Hz noise in the signal containing noise before identifying putative QRS regions in the signal containing noise.

10. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions, when executed by the processor, further causes the processor to perform wavelet filtering on the signal containing noise before identifying putative QRS regions in the signal containing noise.

11. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions causes the processor to identify putative QRS regions in the signal containing noise by filtering the signal, setting a threshold, and identifying spikes above the threshold as putative R-spike components of putative QRS regions.

12. The non-transitory computer-readable storage medium of claim 11, wherein the set of instructions, when executed by the processor, further causes the processor to calculate a putative heart rate (HR) from the R-spikes and to use the HR to modify the signal containing noise to remove baseline wander.

13. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions causes the processor to cross-correlate the putative QRS regions to determine correlated QRS regions of the signal containing noise by cross-correlating each putative QRS region with every other putative QRS region.

14. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions causes the processor to determine correlated QRS regions of the signal containing noise when a peak correlation between a pair of putative QRS regions is above a correlation threshold and their amplitudes vary by less than an amplitude threshold.

15. The non-transitory computer-readable storage medium of claim 14, wherein the correlation coefficient threshold is about 0.8 and the amplitude threshold is about 40%.

16. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions, when executed by the processor, further causes the processor to indicate that there may be a problem with the signal containing noise when there are fewer than a minimum number of correlated consecutive putative QRS regions.

17. The non-transitory computer-readable storage medium of claim 16, wherein the minimum number of correlated consecutive putative QRS regions is 6.

18. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions causes the processor to construct the de-noised ECG signal from the signal by removing the correlated QRS regions from the signal containing noise to form a subtracted signal, filtering the correlated QRS regions of the signal using the first filtering regime and filtering the subtracted signal comprising the remainder of the signal outside of the correlated QRS regions using the second filtering regime, and then adding together the filtered correlated QRS regions with the filtered subtracted signal.

19. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions causes the processor to construct the de-noised ECG signal from the signal by filtering the correlated QRS regions of the signal using the first filtering regime comprising a Principle Component Analysis (PCA).

20. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions causes the processor to construct the de-noised ECG signal from the signal by filtering the remainder of the signal outside of the correlated QRS regions using the second filtering regime comprising a polynomial fit to the remainder of the signal outside of the correlated QRS regions.

21. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions causes the processor to present the de-noised ECG signal by displaying the de-noised ECG signal.

22. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions, when executed by the processor, further causes the processor to determine if the signal is indicative of an atrial fibrillation.

23. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions, when executed by the processor, further causes the processor to indicate if the signal is indicative of an atrial fibrillation by comparing a beat-to-beat variability of the de-noised ECG signal.

24. The non-transitory computer-readable storage medium of claim 23, wherein the set of instructions, when executed by the processor, further causes the processor to determine a QRS onset, QRS offset, and T-wave offset from an average beat.

25. The non-transitory computer-readable storage medium of claim 1, wherein the set of instructions, when executed by the processor, further causes the processor to calculate an average beat by averaging correlated QRS regions that overlap on either side by correlated R-R intervals to form an intermediate average, and then correlating the intermediate average with correlated QRS regions that overlap on either side by correlated R-R intervals, and averaging those correlated QRS regions that overlap on either side by correlated R-R intervals that correlate with the intermediate average by greater than a correlation coefficient threshold of about 0.85, to form the average beat.

26. A non-transitory computer-readable storage medium storing a set of instructions capable of being executed by a processor, that when executed by the processor, causes the processor to extract an electrocardiogram (ECG) signal from a signal containing noise by causing the processor to:

identify putative QRS regions in the signal containing noise;

cross-correlate the putative QRS regions to determine correlated QRS regions of the signal containing noise; and

construct a de-noised ECG signal from the signal containing noise by differentially filtering the correlated QRS regions of the signal containing noise relative to a remainder of the signal containing noise outside of the correlated QRS regions, wherein the constructed de-noised ECG signal comprises the filtered correlated QRS regions of the signal containing noise and the filtered remainder of the signal containing noise outside of the correlated QRS regions;

identify putative R-R intervals in the de-noised ECG signal; and

cross-correlate the putative R-R intervals present the de-noised ECG signal to determine correlated R-R intervals in the de-noised ECG signal; and

modify the de-noised ECG signal by differentially filtering the correlated R-R intervals of the de-noised ECG signal relative to a remainder of the signal containing noise outside of the correlated R-R intervals.

27. A non-transitory computer-readable storage medium storing a set of instructions capable of being executed by a processor, that when executed by the processor, causes the processor to extract an electrocardiogram (ECG) signal from a signal containing noise by causing the processor to:

identify putative QRS regions in a signal containing noise;

cross-correlate each of the putative QRS regions with each other to identify correlated QRS regions;

filter a portion of the signal containing noise corresponding to the correlated QRS regions in isolation using a first filtering regime;

filter a second portion of the signal containing noise corresponding to a region of the signal excluding the correlated QRS regions with a second filtering regime that is different from the first filtering regime;

construct a de-noised ECG signal comprising the filtered portion of the signal containing noise corresponding to the correlated QRS regions and the filtered portion of the signal containing noise corresponding to a region of the signal excluding the correlated QRS regions;

identify putative R-R intervals in the de-noised ECG signal;

cross-correlate the putative R-R intervals to determine correlated R-R intervals;

modify the first ECG signal by filtering only the portion of the de-noised ECG signal corresponding to the correlated R-R intervals with a third filtering regime; and

display the first ECG signal.

28. A system for extracting ECG information from a signal containing noise, the system comprising:

a QRS identifying module configured to receive an electrical signal and to identify a plurality of putative QRS regions in the signal;

a QRS cross-correlator coupled to the QRS identifying module configured to cross-correlate each of the putative QRS regions with each other;

a QRS filter module coupled to the QRS cross-correlator and configured to modify the signal by differentially filtering correlated QRS regions of the signal containing noise relative to other regions of the signal;

an ECG constructing module coupled to the QRS cross-correlator and configured to construct a de-noised ECG comprising the differentially filtered correlated QRS regions of the signal containing noise relative to other regions of the signal;

an R-R cross-correlator adapted to receive the modified signal from the QRS filter module and to cross-correlate putative R-R intervals in the modified signal with each other; and

an R-R filter module coupled to the R-R cross-correlator and configured to further modify the modified signal by differentially filtering correlated R-R intervals relative to other regions of the modified signal.

29. The system of claim 28, further comprising a pre-filtering module connected to the QRS identifying module and configured to pre-filter the received signal containing noise before it is passed to the QRS identifying module.

30. The system of claim 28, further comprising an atrial fibrillation detection module configured to receive the further modified signal from the R-R filter module and to output an indicator if the further modified signal is indicative of atrial fibrillation.

31. The system of claim 28, wherein the QRS identifying module comprises a R-wave detection module having a band pass filter and moving window integrator.

32. A non-transitory computer-readable storage medium storing a set of instructions capable of being executed by a processor, that when executed by the processor, causes the processor to determine if an electrocardiogram (ECG) signal containing noise is indicative of atrial fibrillation, by causing the processor to:

identify putative QRS regions in a signal;

cross-correlate the putative QRS regions to determine correlated QRS regions of the signal;

construct a de-noised ECG signal from the signal containing noise by differentially filtering the correlated QRS regions of the signal containing noise relative to a remainder of the signal outside of the correlated QRS regions, wherein the constructed de-noised ECG signal comprises the filtered correlated QRS regions of the signal containing noise and the filtered remainder of the signal containing noise outside of the correlated QRS regions;

identify putative R-R intervals in the de-noised ECG signal; and

cross-correlate the putative R-R intervals present the de-noised ECG signal to determine correlated R-R intervals in the de-noised ECG signal;

modify the de-noised ECG signal by differentially filtering the correlated R-R intervals of the de-noised ECG signal relative to a remainder of the signal containing noise outside of the correlated R-R intervals; and

determine if the de-noised ECG signal is indicative of atrial fibrillation.

33. The non-transitory computer-readable storage medium of claim 32, wherein the set of instructions causes the processor to determine if the de-noised ECG signal is indicative of atrial fibrillation by examining a number of turning points of correlated R-R intervals in the de-noised ECG signal.

34. The non-transitory computer-readable storage medium of claim 32, wherein the set of instructions causes the processor to determine if the de-noised ECG signal is indicative of atrial fibrillation by measuring beat-to-beat variability, beat-to-every-other-beat variability, and beat-to-every-third-beat variability.

35. The non-transitory computer-readable storage medium of claim 32, wherein the set of instructions causes the processor to determine if the de-noised ECG signal is indicative of atrial fibrillation by measuring beat-to-beat variability, beat-to-every-other-beat variability, and beat-to-every-third-beat variability and by examining the number of turning points of R-R intervals in the de-noised ECG signal.

36. The non-transitory computer-readable storage medium of claim 32, wherein the set of instructions causes the processor to determine if the de-noised ECG signal is indicative of atrial fibrillation by calculating a plurality of predictor beats from a window of n beats in the de-noised ECG signal as the window is moved though the de-noised ECG signal, where n is greater than three.

37. The non-transitory computer-readable storage medium of claim 32, wherein the set of instructions causes the processor to determine if the de-noised ECG signal is indicative of atrial fibrillation by calculating a plurality of predictor beats from a window of n beats in the de-noised ECG signal as the window is moved though the de-noised ECG signal, where n is about 10 or more.

38. The non-transitory computer-readable storage medium of claim 32, wherein the set of instructions causes the processor to determine if the de-noised ECG signal is indicative of atrial fibrillation by calculating a plurality of predictor beats from a window of n beats in the de-noised ECG signal as the window is moved though the de-noised ECG signal, where n is greater than three and by examining a number of turning points of R-R intervals in the de-noised ECG signal.

39. The non-transitory computer-readable storage medium of claim 32, wherein the set of instructions causes the processor to determine if the de-noised ECG signal is indicative of atrial fibrillation by calculating an average beat from the de-noised ECG signal and correlating a proposed P-wave with the region of the average beat before the QRS onset of the average beat.

40. The non-transitory computer-readable storage medium of claim 39, wherein the set of instructions causes the processor to calculate an average beat by averaging correlated QRS regions that overlap on either side by correlated R-R intervals to form an intermediate average, and then correlating the intermediate average with correlated QRS regions that overlap on either side by correlated R-R intervals, and averaging those correlated QRS regions that overlap on either side by correlated R-R intervals that correlate with the intermediate average by greater than a threshold of about 0.85, to form the average beat.

41. A computer implemented method for extracting an electrocardiogram (ECG) signal from a signal containing noise comprising the steps of:

identifying putative ECG sub-regions in a signal;

cross-correlating the putative sub-regions to determine correlated sub-regions of the signal; and

constructing a de-noised ECG signal from the signal containing noise by filtering the correlated sub-regions of the signal using a first filtering regime and filtering a remainder of the signal outside of the correlated sub-regions using a second filtering regime that is different from the first filtering regime, wherein the constructed de-noised ECG comprises the filtered correlated sub-regions of the signal using a first filtering regime and the filtered remainder of the signal outside of the correlated sub-regions.

42. A computer implemented method for extracting an electrocardiogram (ECG) signal from a signal containing noise comprising the steps of:

identifying putative QRS regions in a signal;

cross-correlating the putative QRS regions to determine correlated QRS regions of the signal; and

constructing a de-noised ECG signal from the signal by filtering the correlated QRS regions of the signal using a first filtering regime and filtering a remainder of the signal outside of the correlated QRS regions using a second filtering regime that is different from the first filtering regime, wherein the constructed de-noised ECG comprises the filtered correlated QRS regions of the signal using a first filtering regime and the filtered remainder of the signal outside of the correlated QRS.