Low-distortion ECG denoising

• US 10,213,163 B2
• Filed: 05/07/2018
• Issued: 02/26/2019
• Est. Priority Date: 11/25/2013
• Status: Active Grant

• Current Assignee: Vital Connect, Inc.
• Original Assignee: Vital Connect, Inc.
• Inventors: Ferdosi, Nima, Narasimhan, Ravi
• Primary Examiner(s): Mahmood, Nadia A
• Application Number:
  US15/972,342
• Publication Number:
  US 20180249961A1
• Time in Patent Office: 295 Days
• Field of Search: 600517, 600301
• CPC Class Codes: A61B 5/0006 ECG or EEG signals
• CPC Class Codes: A61B 5/316 Modalities, i.e. specific diagnostic methods
• CPC Class Codes: A61B 5/346 Analysis of electrocardiograms
• CPC Class Codes: A61B 5/352 Detecting R peaks, e.g. for synchronising diagnostic apparatus; Estimating R-R interval
• 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

A method and system for low-distortion denoising of an ECG signal are disclosed. The method comprises determining at least one beat of the ECG signal for denoising using a beat selection logic and denoising the at least one beat using at least one ensemble averaging filter. The system includes a sensor to detect the ECG signal, a processor coupled to the sensor, wherein the processor includes a beat selection logic unit, and a memory device coupled to the processor, wherein the memory device includes an application that, when executed by the processor, causes the processor to determine at least one beat of the ECG signal for denoising using a beat selection logic and to denoise the at least one beat using at least one ensemble averaging filter.

Claims

1. A method for low-distortion denoising of an ECG signal by a wireless sensor device, comprising:

determining, by the wireless sensor device, a plurality of beats of the ECG signal for denoising using the beat selection logic unit;

stretching, by the wireless sensor device, in time the plurality of beats by resampling the plurality of beats of the ECG signal to create a predetermined length for each beat of the plurality of beats;

aligning, by the wireless sensor device, the stretched beats of the plurality of beats by aligning peaks of neighboring beats of the plurality of beats; and

denoising, by the wireless sensor device, the plurality of beats using at least one ensemble averaging filter across the aligned beats of the plurality of beats.

2. The method of claim 1, further comprising:

detecting, by the wireless sensor device, at least one QRS peak of the ECG signal using a narrow bandwidth filter.

3. The method of claim 2, further comprising:

canceling, by the wireless sensor device, baseline wander of the ECG signal using interpolation between values of TP segments of consecutive beats of the ECG signal.

4. The method of claim 3, wherein the cancelling further comprises:

calculating, by the wireless sensor device, a plurality of mid-beats of the ECG signal;

interpolating, by the wireless sensor device, between the plurality of mid-beats to reconstruct a baseline; and

removing, by the wireless sensor device, the baseline from the ECG signal.

5. The method of claim 3, further comprising:

resampling, by the wireless sensor device, a plurality of beats of the ECG signal to create a predetermined length for each beat of the plurality of beats, wherein the resampling time aligns same sections of neighboring beats of the plurality of beats.

6. The method of claim 1, wherein the beat selection logic is used to exclude outlier beats of the ECG signal from the denoising.

7. The method of claim 1, wherein the beat selection logic includes any of polarity change detection, irregular RR interval detection, irregular QRS amplitude detection, statistical change penalties, and posture detection.

8. The method of claim 1, further comprising:

determining, by the wireless sensor device, whether a difference between at least two beats of the ECG signal is due only to noise, wherein the at least two beats are not excluded from the denoising if the difference between the at least two beats is due only to noise.

9. The method of claim 8, wherein the determining further comprises:

calculating, by the wireless sensor device, a Kurtosis of a difference between a latest beat and a reference beat; and

comparing, by the wireless sensor device, the calculated Kurtosis to a Kurtosis threshold.

10. The method of claim 8, wherein the at least one ensemble averaging filter is used to calculate the reference beat.

11. A non-transitory computer-readable medium storing executable instructions that, in response to execution, causes a wireless sensor device to perform operations comprising:

determining, by the wireless sensor device, a plurality of beats of the ECG signal for denoising using the beat selection logic unit;

stretching, by the wireless sensor device, in time the plurality of beats by resampling the plurality of beats of the ECG signal to create a predetermined length for each beat of the plurality of beats;

aligning, by the wireless sensor device, the stretched beats of the plurality of beats by aligning peaks of neighboring beats of the plurality of beats; and

denoising, by the wireless sensor device, the plurality of beats using at least one ensemble averaging filter across the aligned beats of the plurality of beats.

12. The non-transitory computer-readable medium of claim 11, storing further executable instructions that, in response to execution, causes the wireless sensor device to perform further operations comprising:

detecting, by the wireless sensor device, at least one QRS peak of the ECG signal using a narrow bandwidth filter.

13. The non-transitory computer-readable medium of claim 12, storing further executable instructions that, in response to execution, causes the wireless sensor device to perform further operations comprising:

canceling, by the wireless sensor device, baseline wander of the ECG signal using interpolation between values of TP segments of consecutive beats of the ECG signal.

14. The non-transitory computer-readable medium of claim 13, wherein the cancelling further comprises:

calculating, by the wireless sensor device, a plurality of mid-beats of the ECG signal;

interpolating, by the wireless sensor device, between the plurality of mid-beats to reconstruct a baseline; and

removing, by the wireless sensor device, the baseline from the ECG signal.

15. The non-transitory computer-readable medium of claim 13, storing further executable instructions that, in response to execution, causes the wireless sensor device to perform further operations comprising:

resampling, by the wireless sensor device, a plurality of beats of the ECG signal to create a predetermined length for each beat of the plurality of beats, wherein the resampling time aligns same sections of neighboring beats of the plurality of beats.

16. The non-transitory computer-readable medium of claim 11, wherein the beat selection logic is used to exclude outlier beats of the ECG signal from the denoising.

17. The non-transitory computer-readable medium of claim 11, wherein the beat selection logic includes any of polarity change detection, irregular RR interval detection, irregular QRS amplitude detection, statistical change penalties, and posture detection.

18. The non-transitory computer-readable medium of claim 11, storing further executable instructions that, in response to execution, causes the wireless sensor device to perform further operations comprising:

determining, by the wireless sensor device, whether a difference between at least two beats of the ECG signal is due only to noise, wherein the at least two beats are not excluded from the denoising if the difference between the at least two beats is due only to noise.

19. The non-transitory computer-readable medium of claim 18, wherein the determining further comprises:

calculating, by the wireless sensor device, a Kurtosis of a difference between a latest beat and a reference beat; and

comparing, by the wireless sensor device, the calculated Kurtosis to a Kurtosis threshold.

20. The non-transitory computer-readable medium of claim 11, storing further executable instructions that, in response to execution, causes the wireless sensor device to perform further operations comprising:

time-scaling, by the wireless sensor device, the at least one denoised beat back to an original length.