Surface ECG frequency analysis system and method based upon spectral turbulence estimation

• US 5,092,341 A
• Filed: 06/18/1990
• Issued: 03/03/1992
• Est. Priority Date: 06/18/1990
• Status: Expired due to Fees

• Current Assignee: Spacelabs Healthcare LLC (OSI Systems Incorporated)
• Original Assignee: Del Mar Avionics
• Inventors: Kelen, George J.
• Primary Examiner(s): Kamm, William E.
• Application Number:
  US07/539,581
• Time in Patent Office: 624 Days
• Field of Search: 128/696, 128/702, 128/703, 128/704, 128/705, 128/710, 364/413.06
• US Class Current: 600/515
• CPC Class Codes: A61B 5/363 Detecting tachycardia or bradycardia
• CPC Class Codes: A61B 5/7239 using differentiation including higher order derivatives
• CPC Class Codes: A61B 5/7253 characterised by using transforms
• CPC Class Codes: A61B 5/7257 using Fourier transforms

Abstract

Frequency domain ECG signal processing systems and methods plot spectral maps and compute statistical parameters from surface electrocardiographic signals, which plots and parameters reveal abnormalities of electrical conduction within the hearts of patients at risk of ventricular tachycardia. Fourier analysis of short overlapping time segments of QRS ECG signal are used to create three dimensional spectral maps disclosing changes in the frequency spectral content of the ECG signal over the whole QRS region. Correlations and statistical evaluations of the spectral content between QRS time segment pairs quantify the spectral turbulence of the ECG and distinguish those at risk from ventricular tachyarrhythmias.

Claims

1. A method for detecting abnormalities of heart having a time varying QRS signal, said method comprising the steps of

partitioning said QRS signal into a sequence of time displaced time segments,

transforming said time segments into respective sets of spectral components, and

correlating said sets of spectral components across the QRS signal, and

generating an abnormality score when the correlation of said correlating step is above a predetermined state.

2. The method of claim 1 wherein said transforming step comprises the step of

partitioning each of said time segment into data points, and

executing a Fourier Transform upon the data points.

3. The method of claim 1 wherein said transforming step comprises the step of

filtering said QRS signal into spectral components, and

isolating said spectral components by time segment.

4. The method of claim 1 further comprising the step of signal averaging said QRS signal prior to said partitioning step.

5. The method of claim 1 further comprising the step of generating spectral turbulence parameters based upon the degree of correlation of said sets of spectral components.

6. The method of claim 1 further comprising the step of differentiating said QRS signal prior to said transforming step.

7. The method of claim 1 wherein said correlating step comprising the steps of

correlating successive pairs of sets of spectral components of respective successive time segment pairs, generating an intersegment correlation mean, and assigning an abnormality value when said intersegment correlation mean is above a predetermined value.

8. The method of claim 1 wherein said correlating step comprising the steps of

correlating successive pairs of sets of spectral components of respective successive time segment pairs,

generating an intersegment standard deviation, and assigning an abnormality value when said intersegment correlation standard deviation is above a predetermined value.

9. The method of claim 1 wherein said correlating step comprising the steps of

correlating successive pairs of sets of spectral components of respective successive time segment pairs,

generating an intersegment correlation ratio based upon the percentage of intersegment correlations is below another predetermined value, and

assigning an abnormality value when said intersegment correlation mean is above a predetermined value.

10. The method of claim 1 wherein said correlating step comprising the steps of

computing an average set of spectral components,

correlating each of said sets of spectral components of said respective time segments to said average set of spectral components,

generating a spectral entropy value based upon said correlation of said average set of spectral components, and

assigning an abnormality value when said spectral entropy is above a predetermined value.

11. The method of claim 1 wherein said generating step comprises the step of,

computing an abnormality score based upon equal weight of intersegment correlation mean, intersegment correlation standard deviation, intersegment correlation ratio, and spectral entropy spectral turbulence correlations over the QRS region, and

displaying said abnormality score.

12. A method for detecting abnormalities of heart having a time varying QRS signal, said method comprising the steps of

signal averaging an ECG signal

identifying the QRS signal of said ECG signal,

differentiating said QRS signal

dividing said differentiated QRS signal into a sequence of overlapping time segments,

dividing each of said time segment into data points,

multiplying data points of each of said time segments by a window function to weight each time segment,

performing a discrete fast Fourier transform on each said windowed time segments, in which the power spectral density of a series of frequency components are calculated, and

producing a numerical matrix or table ofpower spectral densities are located by frequency components and time segments,

correlating said power spectral densities of said frequency components with each other across the QRS signal,

computing an spectral turbulence parameters based upon the correlation of said power spectral densities,

assigning an abnormality state in the presence of abnormal spectral turbulence.

13. The method of claim 12 further including locating the onset and offset of the QRS portion of said ECG signal.

14. The method of claim 12 wherein said spectral turbulence parameters comprise intersegment correlation mean, intersegment correlation standard deviation, intersegment correlation ratio, and spectral entropy.

15. The method of claim 12 wherein said abnormality state is determined by an abnormality score above a predetermined value, said abnormality score incremented for each of said spectral turbulence parameters in an abnormal state.

16. The method of claim 12 wherein

said time segments are between twelve and forty-eight milli seconds,

said time segments are stepped between one and five milli seconds,

said data points are one milli second apart, and

said transform uses a matrix having points equals to at least the number of said data points of said time segment and equals a number of the power of two.

17. A method for detecting abnormalities of heart having a time varying QRS signal, said method comprising the steps of

signal averaging an ECG signal

identifying the QRS signal of said ECG signal,

differentiating said QRS signal

dividing said differentiated QRS signal into a sequence of overlapping time segments,

multiplying each said time segment by a window function to weight each time segment,

performing a discrete fast Fourier transform on each said windowed time segments, in which the power spectral density of a series of frequency components are calculated,

producing a numerical matrix or table ofpower spectral densities are located by frequency components and time segments,

correlating said power spectral densities of said frequency components with each other across the QRS signal,

computing an spectral turbulence parameters based upon the correlation of said power spectral densities,

assigning an abnormality state in the presence of abnormal spectral turbulence.

18. The method of claim 17 further including locating the onset and offset of the QRS portion of said ECG signal.

19. The method of claim 17 wherein said spectral turbulence parameters comprise intersegment correlation mean, intersegment correlation standard deviation, intersegment correlation ratio, and spectral entropy.

20. The method of claim 17 wherein said abnormality state is determined by an abnormality score above a predetermined value, said abnormality score incremented for each of said spectral turbulence parameters in an abnormal state.

21. The method of claim 20 wherein said computing step further comprises the step of

computing each of said spectral turbulence parameters for each of an X, Y and Z lead, and

averaging each of said spectral turbulence parameter on said computing Z, Y and Z lead whereby said abnormality score is incremented based upon an average of each of said spectral turbulence parameters in an abnormal state.

22. A system for detecting heart conditions comprising

first means for receiving said a QRS signal, and

second means for generating an abnormality indicator, said second means connected to said first means, for dividing said QRS signal into a sequence of overlapping time segments each have a plurality of data points, for transforming said data points of each of said time segments into a matrix ofpower content indicators organized by time segment and frequency components, for correlating said power content indicators of said frequency components with said correlation upon successive time segments, for generating at least one spectral turbulence parameter based upon the correlation of successive time segments with said abnormality indicator based upon the value of said at least one spectral turbulence parameter.

23. The system of claim 22 wherein said QRS signal is a time average QRS signal and wherein said second means identifies an onset and an offset of said QRS time averaged signal.

24. The system of claim 22 wherein said at least one spectral turbulence parameter is an intersegment correlation mean.

25. The system of claim 22 wherein said at least one spectral turbulence parameter is an intersegment correlation standard deviation.

26. The system of claim 22 wherein said at least one spectral turbulence parameter is an intersegment correlation ratio.

27. The system of claim 22 wherein said at least one spectral turbulence parameter is a spectral entropy value.

28. A system for detecting heart conditions comprising

first means for storing a sequence of QRS signals from a human heart,

second means for receiving said QRS signals form said first means,

third means for generating at least one abnormality indication, said third means connected to said second means, said third means for signal averaging said QRS signals, for dividing said signal averaged QRS signal into a sequence of overlapping time segments each have a plurality of data points, for transforming said data points of each of said time segments into a matrix ofpower content indicators organized by time segment and frequency components, for correlating said power content indicators of said frequency components with said correlation upon successive time segments, for generating at least one spectral turbulence parameter based upon the correlation of successive time segments with said at least one abnormality indication based upon the value of said at least one spectral turbulence parameter.

29. A system for detecting heart conditions comprising

first means for receiving at least one QRS signal,

second means for generating at least one abnormality indication, said second means connected to said first means, said second for dividing said at least one QRS signal into a sequence of overlapping time segments each have a plurality of data points, for transforming said data points of each of said time segments into a matrix ofpower content indicators organized by time segment and frequency components, for correlating said power content indicators of said frequency components with said correlation upon successive time segments, for generating at least one spectral turbulence parameter based upon the correlation of successive time segments with said at least one abnormality indication based upon the value of said at least one spectral turbulence parameter, and

third means for displaying said at least one abnormality indicator.

30. A system for detecting heart conditions comprising

a scanner adapted to receive Holter tapes storing a sequence of QRS signals, said scanner for scanning said tapes at high speed,

a digital to analog converter for converting said QRS signal into digital type signal suitable for storing in a digital computer having memory,

digital computer for generating at least one abnormality indication, said digital computer connected to a memory storing said converted QRS signal, said digital computer for dividing said at least one QRS signal into a sequence of overlapping time segments each have a plurality of data points, for transforming said data points of each of said time segments into a matrix ofpower content indicators organized by time segment and frequency components, for correlating said power content indicators of said frequency components with said correlation upon successive time segments, for generating at least one spectral turbulence parameter based upon the correlation of successive time segments with said at least one abnormality indication based upon the value of said at least one spectral turbulence parameter.

31. The system of claim 30 further comprising display means for displaying to an operator said at least one abnormality indicator.

32. The system of claim 30 further comprising

Holter recorder means worn by a patient to record the patient'"'"'s QRS signal on Holter magnetic tape, and

Holter magnetic tape suitable for storing analog QRS signal over an ambulatory period.

33. The system of claim 30 wherein said display means is a CRT connected to said digital computer.

34. The system of claim 30 wherein said display means is a printer connected to said digital computer.

35. An apparatus for providing an indicator of electrical heart conduction, said apparatus comprising

means for receiving QRS signal from a human being,

electronic filters for isolating frequency components of said QRS signal,

comparators for comparing isolated frequency components at intervals over said QRS signal, and

electronic circuit connected to comparators for generating an indication of spectral turbulence.