Robust time-diversity filter and method for removing electromagnetic interference

• US 5,687,735 A
• Filed: 03/28/1996
• Issued: 11/18/1997
• Est. Priority Date: 03/28/1996
• Status: Expired due to Fees

• Current Assignee: Agilent Technologies, Inc.
• Original Assignee: Hewlett-Packard Company (HP Inc.)
• Inventors: Forbes, Alfred Dean, Helfenbein, Eric D.
• Primary Examiner(s): Getzow, Scott M.
• Application Number:
  US08/624,194
• Time in Patent Office: 600 Days
• Field of Search: 128/696, 128/901, 364/724.01, 364/572, 364/573, 364/574
• US Class Current: 600/509
• CPC Class Codes: A61B 5/276 Protection against electrode failure
• CPC Class Codes: A61B 5/7203 for noise prevention, reduction or removal
• CPC Class Codes: Y10S 128/901 Suppression of noise in electric signal

Abstract

An apparatus and method for the filtering of signals collected during a period that includes a quiet ("signal-free") epoch and an event epoch is disclosed. The apparatus includes a means for generating electrical digitized signals of the event episode, memory for receiving and storing the digitized signals of the event episode, an event detector for determining the event epoch and the quiet epoch of the event episode in the digitized signals, a spectrum analyzer to analyze the power spectrum of the digitized signals of at least one event episode, an estimator to estimate electromagnetic interference (EMI) based on the digitized signals of the quiet epoch and the frequencies obtained in the spectral analysis; and a subtractor to subtract the estimated EMI from the digitized signals of the event epoch. The resulting signals are signals of the event epoch with reduced EMI. The apparatus uses basis function(s) at one or more dominant frequencies determined from the spectral analysis to curve-fit data of the quiet-epoch signals to generate the EMI estimate.

Claims

1. Apparatus for filtering signals of a series of generally periodic events each having an event episode including an event epoch and a quiet epoch, the apparatus comprising:

(a) means for generating electrical digitized signals of the event episodes;

(b) memory connected for receiving and storing the digitized signals of the event episodes;

(c) event detector connected to determine the event epoch and the quiet epochs of the event episodes for the events in the digitized signals;

(d) spectrum analyzer connected to analyze power spectra of the digitized signals, each spectrum being based on at least one event episode of the events at a time;

(e) estimator connected to estimate electromagnetic interference (EMI) by comparing the digitized signals of a quiet epoch of the at least one event episode with basis function at one or more frequencies identified in a spectrum in the spectral analysis by the spectrum analyzer to obtain an EMI estimate;

such that the estimator by said spectra can obtain EMI estimates adapted to a shift in frequency phase or amplitude in the EMI for EMI that varies with time; and

(f) subtractor connected to subtract from the digitized signals of the event epochs the EMI estimates derived from corresponding spectra such that the resulting signals for the event epochs have reduced EMI.

2. The apparatus according to claim 1 wherein the estimator is adapted to combine a sine wave and a cosine wave of an EMI frequency determined from the spectral analysis to fit data of the signals of the quiet epoch to generate the EMI estimate.

3. The apparatus according to claim 2 wherein only one sine wave and only one cosine wave of the EMI frequency are combined to fit data of the quiet epoch at the EMI frequency, said sine wave and said cosine wave can have different amplitudes, each of said amplitudes being selected such that the combination of said sine wave and said cosine wave generates the EMI estimate at the EMI frequency.

4. The apparatus according to claim 2 wherein the estimator is adapted to generate an EMI estimate having a fundamental frequency and one or more of its harmonic frequencies.

5. The apparatus according to claim 2 wherein the estimator is adapted to estimate the EMI by combining a sine wave and a cosine wave of an EMI frequency determined from combined data including data of signals from at least two quiet epochs.

6. The apparatus according to claim 1 wherein the memory, event detector, estimator, and subtractor are part of an electronic computer.

7. The apparatus according to claim 1 wherein the means for generating electrical digitized signals comprises an analog-to-digital converter connectable to a sensor for collecting the signals of the event episodes from a signal source.

8. The apparatus according to claim 1 wherein the means for generating electrical digitized signals comprises a sensor adapted to collect signals of the event episodes from a signal source.

9. The apparatus according to claim 8 wherein the sensor comprises electrocardiogram (ECG) electrodes for attaching to a patient.

10. The apparatus according to claim 8 further comprising an EMI-signal-collector to be positioned on the patient at a location different from that of the sensor such that the EMI-signal-collector'"'"'s EMI signals during the quiet epoch of an event can be compared with its EMI signals collected during the event epoch of the event so that event epochs during which unusual EMI signals occur can be discarded.

11. The apparatus according to claim 1 further comprising a high-pass filter connected to suppress non-EMI signals from the quiet epoch in an event before the estimator and wherein the estimator scales up the estimate to compensate for attenuation of the EMI in the quiet-epoch signals by the high-pass filter.

12. The apparatus according to claim 1 wherein the estimator comprises a robust means for estimating EMI such that the effect of outlier signals is reduced.

13. The apparatus according to claim 1 wherein the estimator compares blocks of data of signals of the quiet epochs to blocks of data of basis functions for estimating EMI for the event episodes.

14. The apparatus according to claim 1 wherein the estimator by means of the spectra can adapt to a shift in EMI frequency.

15. Apparatus for filtering electrocardiogram (ECG) signals of a series of heart beats each having a cardiac event episode including an event epoch suspected of containing signals of interest and a quiet epoch suspected to be free of said signals of interest, the apparatus comprising:

(a) detector for detecting analog ECG signals from a patient;

(b) analog-to-digital converter connected for converting the analog ECG signals to digitized signals; and

(c) computer connected for receiving and storing the digitized signals of the event episode, for determining the event epochs and the quiet epochs of the event episodes in the digitized signals of the series of the art beats, for analyzing power spectra of the digitized signals, each spectrum being based on the signals of at least one event episode, for estimating electromagnetic interference (EMI) to determine an EMI estimate for each of the event epochs based on the frequencies obtained in the spectrum corresponding to said each of the event epochs such that the estimator by means of the spectra can adapt to a shift in frequency in the EMI for EMI that varies in frequency with time; and

for subtracting the EMI estimates from the digitized signals of the corresponding event epoch such that the resulting signals for the event epochs have reduced EMI, said estimating for each event epoch comprises filtering the signals of a quiet epoch of the at least one event episode with a high-pass filter to supress non-EMI signals in the signals of the quiet epoch, combining a sine wave and a cosine wave of an EMI frequency determined from the spectral analysis to approximate data of the signals of the quiet epoch after non-EMI-signal-suppression such that a block of data of the signals of the quiet epoch is compared to a block of data of sine and cosine waves to generate a first EMI estimate, and scaling up the first EMI estimate to compensate for attenuation of EMI in the signals of the quiet epoch by the high-pass filter to arrive at the EMI estimate for the subtraction.

16. A method for filtering signals of a series of generally periodic events each having an event episode including an event epoch and a quiet epoch, the method comprising:

(a) detecting signals of the event epochs and quiet epochs of the generally periodic events;

(b) analyzing power spectra, each spectrum being based on signals of at least one of the event episodes;

(c) estimating electromagnetic interference (EMI) of the event epochs, said estimating from each event including matching signals of at least one quiet epoch of the at least one of the event episodes to basis functions of one or more frequencies identified in a spectrum corresponding to said at least one of the event episodes, such that the estimating step can adapt to a shift in frequency, phase, or amplitude in the EMI for EMI that varies with time; and

(d) subtracting the corresponding estimated EMI from the signals of the event epoch such that the resulting signals are signals of the event epochs with reduced EMI.

17. The method according to claim 16 wherein the signals are digitized signals and wherein estimating EMI for an event epoch comprises combining a sine wave and a cosine wave of an EMI frequency determined from the spectral analysis to fit data of the signals of the quiet epoch corresponding to said event epoch.

18. The method according to claim 17 wherein only one sine wave and only one cosine wave of the EMI frequency are combined to fit data of the quiet epoch at the EMI frequency, said sine wave and said cosine wave can have different amplitudes, each of said amplitudes being selected such that the combination of said sine wave and said cosine wave generates the estimated EMI at the EMI frequency.

19. The method according to claim 17 wherein estimating EMI generates an EMI estimate having a fundamental frequency and one or more harmonic frequencies of said fundamental frequency.

20. The method according to claim 17 wherein estimating EMI comprises combining a sine wave and a cosine wave of an EMI frequency which is determined from combined data including data of signals of at least two quiet epochs.

21. The method according to claim 16 further comprising converting analog signals to digital signals before estimating EMI.

22. The method according to claim 16 further comprising collecting signals from a signal source.

23. The method according to claim 16 further comprising collecting electrocardiogram (ECG) signals of a patient for filtering of EMI.

24. The method according to claim 16 further comprising collecting signals of the event episodes for filtering of EMI and collecting EMI signals at a location different from where the signals of the event episodes are collected so that EMI signals collected thereby from said different location during a quiet epoch can be compared with EMI signals collected during the corresponding event epoch so that event epochs during which unusual EMI signals occur can be discarded.

25. A program storage medium readably by a computer, tangibly embodying a program of instruction including code means executable by the computer to digitally filter signals of a series of events each having an event episode including an event epoch and a quiet epoch, the program storage medium comprising:

a memory device having code means including,

(a) code means for detecting the event epochs and the quiet epochs for the series prevents;

(b) code means for analyzing power spectra of digitized signals, each spectrum being based on at least one event episodes of the events;

(c) code means for estimating electromagnetic interference (EMI) of the event epoch, for each event epoch said code means for estimating matches signals of a quiet epoch of the at least one event episode of a corresponding spectrum to a combination of a sine wave and a cosine wave at an EMI frequency determined from the spectrum to fit data of the signals of the quiet epoch to arrive at a first estimate, said estimating code means including a high-pass-filter for suppressing non-EMI signals from the signals of the quiet epoch and a means to scale up the first estimate obtained by said matching to compensate for attenuation of EMI the signals of the quiet epoch by the high-pass-filter to arrive at the EMI estimate, such that said code means for estimating can adapt to a shift in frequency in the EMI for EMI that varies in frequency with time; and

(d) code means for subtracting the corresponding estimated EMI from the signals of the event epochs such that the resulting signals are signals of the event epochs with reduced EMI.