Electrocardiograph (ECG) signal processing

• US 9,629,564 B2
• Filed: 09/26/2014
• Issued: 04/25/2017
• Est. Priority Date: 09/26/2014
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Jonnada, Sowmya, Natarajan, Venkat, Baxi, Amit
• Primary Examiner(s): Stice, Paula J
• Application Number:
  US14/498,060
• Publication Number:
  US 20160089047A1
• Time in Patent Office: 942 Days
• Field of Search: 600516
• CPC Class Codes: A61B 5/316 Modalities, i.e. specific diagnostic methods
• CPC Class Codes: A61B 5/333 Recording apparatus specially adapted therefor
• CPC Class Codes: A61B 5/339 Displays specially adapted therefor
• CPC Class Codes: A61B 5/349 Detecting specific parameters of the electrocardiograph cycle
• CPC Class Codes: A61B 5/352 Detecting R peaks, e.g. for synchronising diagnostic apparatus; Estimating R-R interval
• 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 5/6801 specially adapted to be attached to or worn on the body surface
• CPC Class Codes: A61B 5/6802 Sensor mounted on worn items
• CPC Class Codes: A61B 5/7203 for noise prevention, reduction or removal
• CPC Class Codes: A61B 5/7239 using differentiation including higher order derivatives

Abstract

Technology for processing an electrocardiograph (ECG) signal is disclosed. The ECG signal can be identified, wherein the ECG signal is affected by baseline wander noise. Signal processing can be performed on the ECG signal affected by baseline wander noise in order to determine a start time and an end time for individual waveforms in the ECG signal affected by baseline wander noise. Features for the individual waveforms in the ECG signal can be extracted, wherein the features indicate one or more cardiac function metrics.

Claims

1. A wearable device operable to determine cardiac function metrics of a subject, the wearable device comprising circuitry configured to:

measure an electrocardiograph (ECG) signal using a sensor at the wearable device, wherein the ECG signal includes a noisy ECG signal;

remove a moving average of the noisy ECG from the noisy ECG signal in order to obtain a corrected ECG signal;

determine a plurality of R peak locations in the corrected ECG signal;

identify R peak locations for individual waveforms in the noisy ECG signal using the determined plurality of R peak locations in the corrected ECG signal;

extract features from the noisy ECG signal using the R peak locations for individual waveforms in the noisy ECG signal; and

identify one or more cardiac function metrics based on the features extracted from the noisy ECG signal.

2. The wearable device of claim 1, further configured to provide the cardiac function metrics for display on the wearable device.

3. The wearable device of claim 1, wherein the noisy ECG signal is affected by baseline wander noise.

4. The wearable device of claim 1, wherein the one or more cardiac function metrics indicate at least one of a healthy ECG signal for the subject attached to the wearable device or an unhealthy ECG signal for the subject attached to the wearable device.

5. The wearable device of claim 1, wherein the features extracted from the individual waveforms in the noisy ECG signal include wave amplitudes and time intervals for the individual waveforms in the noisy ECG signal.

6. The wearable device of claim 1, wherein the features extracted from the individual waveforms in the noisy ECG include at least one of:

a PR interval, a QRS width, a QT interval, an ST segment, or an R-R interval.

7. The wearable device of claim 1, wherein the circuitry is further configured to:

calculate the moving average for the noisy ECG signal;

remove the moving average from the noisy ECG signal in order to obtain the corrected ECG signal with noise below a defined threshold, wherein removing the moving average from the noisy ECG signal comprises removing a direct current (DC) offset from the noisy ECG signal; and

determine the plurality of R peak locations in the corrected ECG signal using a defined threshold.

8. The wearable device of claim 7, wherein the circuitry is further configured to:

select an R peak location in an individual waveform in the noisy ECG signal using the plurality of R peak locations that are determined in the corrected ECG signal with noise below the defined threshold;

identify a P peak location for a P peak and a T peak location for a T peak in the individual waveform in the noisy ECG signal using the R peak location and local maxima on the noisy ECG signal;

identify a Q wave location and an S wave location in the individual waveform in the noisy ECG signal using local minima on the noisy ECG signal;

determine an onset value for the P peak and an offset value for the P peak using slope changes in the noisy ECG signal;

determine an onset value for the T peak and an offset value for the T peak using slope changes in the noisy ECG signal;

select a local reference point in the individual waveform in the noisy ECG signal using an amplitude associated with one of;

the onset value for the P peak, the offset value for the P peak, the onset value for the T peak, or the offset value for the T peak; and

extract features from the individual waveform in the noisy ECG signal with respect to the local reference point, wherein the features are used to identify the cardiac function metrics.

9. The wearable device of claim 8, wherein the circuitry is further configured to:

identify one or more slope changes in the individual waveform in the noisy ECG signal; and

identify the P peak location and the T peak location in the individual waveform in the noisy ECG signal using the one or more slope changes.

10. A system operable to detect cardiac function metrics in a subject using signal features, the system comprising:

a wearable sensor configured to receive an electrocardiograph (ECG) signal from the subject, wherein the ECG signal includes a noisy ECG signal;

a feature extraction module configured to;

remove a moving average of the noisy ECG from the noisy ECG signal in order to obtain a corrected ECG signal;

determine a plurality of R peak locations in the corrected ECG signal;

identify R peak locations for individual waveforms in the noisy ECG signal using the determined plurality of R peak locations in the corrected ECG signal; and

extract features from the noisy ECG signal using the R peak locations for individual waveforms in the noisy ECG signal;

a cardiac function metrics module configured to identify one or more cardiac function metrics based on the features extracted from noisy ECG signal; and

a display configured to receive the cardiac function metrics and display the cardiac function metrics.

11. The system of claim 10, wherein at least one of:

the feature extraction module, the cardiac function metrics module, or the display is located external to the wearable sensor.

12. The system of claim 10, wherein the noisy ECG signal is affected by baseline wander noise.

13. The system of claim 10, wherein the features extracted from the individual waveforms in the noisy ECG signal include wave amplitudes and time intervals for the individual waveforms in the noisy ECG signal.

14. The system of claim 10, wherein the features extracted from the individual waveforms in the noisy ECG include at least one of:

a PR interval, a QRS width, a QT interval, an ST segment, or an R-R interval.

15. The system of claim 10, wherein the feature extraction module is further configured to:

calculate the moving average for the noisy ECG signal;

remove the moving average from the noisy ECG signal in order to obtain the corrected ECG signal with noise below a defined threshold, wherein removing the moving average from the noisy ECG signal comprises removing a direct current (DC) offset from the noisy ECG signal; and

determine the plurality of R peak locations in the corrected ECG signal using a defined threshold.

16. The system of claim 15, wherein the feature extraction module is further configured to:

select an R peak location in an individual waveform in the noisy ECG signal using the plurality of R peak locations that are determined in the corrected ECG signal with noise below the defined threshold;

identify a P peak location for a P peak and a T peak location for a T peak in the individual waveform in the noisy ECG signal using the R peak location and local maxima on the noisy ECG signal;

identify a Q wave location and an S wave location in the individual waveform in the noisy ECG signal using local minima on the noisy ECG signal;

determine an onset value for the P peak and an offset value for the P peak using slope changes in the noisy ECG signal;

determine an onset value for the T peak and an offset value for the T peak using slope changes in the noisy ECG signal;

select a local reference point in the individual waveform in the noisy ECG signal using an amplitude associated with one of;

the onset value for the P peak, the offset value for the P peak, the onset value for the T peak, or the offset value for the T peak; and

extract features from the individual waveform in the noisy ECG signal with respect to the local reference point, wherein the features are used to identify the one or more cardiac function metrics.

17. A method for processing an electrocardiograph (ECG) signal, using one or more computer systems configured with executable instructions, the method comprising:

identifying the ECG signal, wherein the ECG signal is affected by baseline wander noise, using a processor;

removing a moving average from the ECG signal affected by the baseline wander noise in order to obtain a corrected ECG signal, using the processor;

determining a plurality of R peak locations in the corrected ECG signal, using the processor; and

selecting an R peak location in an individual waveform in the ECG signal affected by baseline wander noise using the plurality of R peak locations in the corrected ECG signal, using the processor.

18. The method of claim 17, further comprising measuring the ECG signal using a sensor in a wearable device that is attached to a user.

19. The method of claim 17, further comprising determining a cardiac function metrics to indicate at least one of a healthy ECG signal or an unhealthy ECG signal based on features extracted in the individual waveforms in the ECG signal.

20. The method of claim 17, wherein performing the signal processing on the ECG signal affected by baseline wander further comprises:

calculating the moving average for the ECG signal affected by baseline wander noise;

removing the moving average from the ECG signal affected by baseline wander noise in order to obtain the corrected ECG signal with noise below a defined threshold; and

determining the plurality of R peak locations in the corrected ECG signal using a defined threshold.

21. The method of claim 20, further comprising:

identifying a P peak location for a P peak and a T peak location for a T peak in the individual waveform in the ECG signal affected by baseline wander noise using the R peak location and local maxima on the noisy ECG signal;

identifying a Q wave location and an S wave location in the individual waveform in the noisy ECG signal using local minima on the noisy ECG signal;

determining an onset value for the P peak and an offset value for the P peak using slope changes in the noisy ECG signal;

determining an onset value for the T peak and an offset value for the T peak using slope changes in the noisy ECG signal;

selecting a local reference point in the individual waveform in the ECG signal affected by baseline wander noise using an amplitude associated with one of;

the onset value for the P peak, the offset value for the P peak, the onset value for the T peak, or the offset value for the T peak; and

extracting features from the individual waveform in the ECG signal affected by baseline wander noise with respect to the local reference point, wherein the features are used to identify the one or more cardiac function metrics.

22. At least one non-transitory machine readable storage medium comprising a plurality of instructions adapted to be executed to implement the method of claim 17.