Characterizing the statistical properties of a biological signal

• US 4,979,110 A
• Filed: 09/22/1988
• Issued: 12/18/1990
• Est. Priority Date: 09/22/1988
• Status: Expired due to Term

• Current Assignee: Massachusetts Institute of Technology
• Original Assignee: Massachusetts Institute of Technology
• Inventors: Albrecht, Paul, Cohen, Richard J.
• Primary Examiner(s): NOT, DEFINED
• Assistant Examiner(s): NOT, DEFINED
• Application Number:
  US07/247,738
• Time in Patent Office: 817 Days
• Field of Search: 364/413.03, 364/413.05, 364/413.06.728.07, 18/670, 18/671, 18/672, 18/700, 18/702
• US Class Current: 600/301
• CPC Class Codes: A61B 5/021 Measuring pressure in heart or blood vessels A61B5/0205 takes precedence
• CPC Class Codes: A61B 5/024 Detecting, measuring or recording pulse rate or heart rate A61B5/0205, A61B5/021 take precedence
• CPC Class Codes: A61B 5/4035 Evaluating the autonomic nervous system
• CPC Class Codes: A61B 5/7257 using Fourier transforms
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04

Abstract

The statistical properties of a biological signal intermittently obscured by a relatively independent biological process are estimated by noting the time intervals during which the obscuring process is occurring. From the time intervals, a windowing function is constructed which makes a transition from one towards zero as the process obscures the signal and a transition towards one as the process terminates. Thereafter, a statistical characterization is performed in which the windowing function is used to weight the relative contributions of corresponding segments of the biological signal. In preferred embodiments, the statistical properties include the autocorrelation function and power spectrum of biological signals such as heart rate and blood pressure. The invention is applicable also to estimating the crosscorrelation and transfer function of a first signal intermittently obscured by a first process and a second signal intermittently obscured by a second process in which each of the signals are relatively independent of each of the obscuring processes. When the biological signal is heart rate, the obscuring process may be the occurrence of ectopic beats. Intervals which contain ectopic beats are treated as though they are missing. In this way, data which might otherwise be unusable by virtue of ectopic beats can become useful for assessing the status of the biological system.

Claims

1. Method for characterizing the fluctuations of a signal representing an underlying biological process, said signal being intermittently obscured by a relatively independent process resulting in a corrupted signal comprising:

noting the time intervals during which the obscuring process is occurring,

constructing from the intervals a windowing function which makes a transition from one towards zero as the process obscures the signal and a transition towards one as the process terminates, and

using the windowing function to weight the relative contributions of corresponding segments of the corrupted signal to create a new signal,

whereby a characterization of the statistical properties of the new signal can be made to approximate the characterization of the statistical properties of the signal in the absence of the effect of the obscuring process.

2. The method of claim 1 wherein the weighting comprises computing the product of the windowing function and the corrupted signal to create the new signal, and wherein the fluctuations are characterized by an estimate of the autocorrelation function of the signal in the absence of the effect of the obscuring process comprising:

computing the autocorrelation of the new signal,

computing the autocorrelation of the windowing function, and

computing the quotient of the autocorrelation function of the new signal and the autocorrelation of the windowing function.

3. The method of claim 2 further comprising computing a power spectrum estimate from the estimate of the autocorrelation function.

4. The method of claims 1, 2, or 3 wherein the signal is heart rate.

5. The method of claims 1, 2, or 3 wherein the obscuring process is atrial or ventricular ectopic beats.

6. The method of claims 1, 2, or 3 wherein the signal is arterial blood pressure.

7. The method of claims 1, 2, or 3 wherein the signal is cardiac output.

8. Method for characterizing the joint fluctuations between a first signal representing a first underlying biological process, said first signal being intermittently obscured by a first process resulting in a first corrupted signal, and a second signal representing a second underlying biological process, said second signal being intermittently obscured by a second process (which may or may not be the same as the first obscuring process) resulting in a second corrupted signal in which each of the first and second signals are relatively independent of each of the first and second obscuring processes comprising:

noting the time intervals during which the first obscuring process is occurring,

constructing from the intervals a first windowing function which makes a transition from one towards zero as the first process obscures the first signal and a transition towards one as the first process terminates,

noting the time intervals during which the second obscuring process is occurring,

constructing from the second set of time intervals a second windowing function which makes a transition from one towards zero as the second process obscures the second signal and a transition towards one as the second process terminates,

computing the product of the first corrupted signal with the first windowing function to create a new first signal,

computing the product of the second corrupted signal with the second windowing function to create a new second signal,

computing the crosscorrelation of the two new signals,

computing the crosscorrelation of the two windowing functions, and

computing the quotient of the crosscorrelation of the new signals and the crosscorrelation of the windowing functions to produce an estimate of the crosscorrelation function between the two signals in the absence of the effects of the obscuring processes.

9. The method of claim 8 further comprising computing an estimate of the cross-spectrum of the first and second signals in the absence of the effects of the obscuring processes from the estimate of the crosscorrelation function.

10. Method for characterizing the fluctuations between a first signal representing a first underlying biological process, said first signal being intermittently obscured by a first process resulting in a first corrupted signal, and a second signal representing a second underlying biological process, said second signal being intermittently obscured by a second process (which may or may not be the same as first obscuring process) resulting in a second corrupted signal in which each of the first and second signals are relatively independent of each of the first and second obscuring processes comprising:

noting the time intervals during which the first obscuring process is occurring,

constructing from the intervals a first windowing function which makes a transition from one towards zero as the first process obscures the first signal and a transition towards one as the first process terminates,

computing the product of the first corrupted signal and the first windowing function to create a new first signal,

computing the autocorrelation of the new first signal,

computing the autocorrelation of the first windowing function,

computing the quotient of the autocorrelation of the new first signal and the autocorrelation of the first windowing function to produce an estimate of the autocorrelation function of the first signal in the absence of the effect of the first obscuring process,

computing an estimate of the power spectrum of the first signal in the absence of the effect of the first obscuring process from the estimate of the autocorrelation function,

noting the time intervals during which the second obscuring process is occurring,

constructing from the second set of time intervals a second windowing function which makes a transition from one towards zero as the second process obscures the second signal and a transition towards one as the second process terminates,

computing the product of the second corrupted signal with the second windowing function to create a new second signal,

computing the crosscorrelation of the two new signals,

computing the crosscorrelation of the two windowing functions,

computing the quotient of the crosscorrelation of the new signals and the crosscorrelation of the windowing functions to generate an estimate of the crosscorrelation function of the two signals in the absence of the effects of the obscuring processes,

computing from the estimate of the crosscorrelation function an estimate of the cross-spectrum of the two signals in the absence of the effects of the obscuring process, and

computing the quotient of the estimates of the cross-spectrum and the power spectrum to generate an estimate of the transfer function between the two signals in the absence of the effects of the obscuring processes.

11. The method of claim 10 further comprising computing the inverse Fourier transform of the estimate of the transfer function to generate an estimate of the impulse response function between the first and second signals in the absence of the effects of the obscuring processes.

15. The method of any of claims 8-14 wherein one of the signals is heart rate.

16. The method of any of claims 8-14 wherein one of the obscuring processes is atrial or ventricular ectopic beats.

17. The method of any of claim 8-14 wherein one of the signals is aterial blood pressure.

18. The method of any of claims 8-14 wherein one of the signals is cardiac output.

12. Method for characterizing the fluctuations between a first signal representing a first underlying biological process, said first signal being intermittently obscured by a first process resulting in a first corrupted signal, and a second signal representing a second underlying biological process, said second signal being intermittently obscured by a second process (which may or may not be the same as the first obscuring process) resulting in a second corrupted signal in which each of the first and second signals are relatively independent of each of the first and second obscuring processes comprising:

noting the time intervals during which the first obscuring process is occurring,

constructing from the intervals a first windowing function which makes a transition from one towards zero as the first process obscures the first signal and a transition towards one as the first process terminates,

computing the product of the first corrupted signal and the first windowing function to create a new first signal,

computing the autocorrelation of the new first signal,

computing the autocorrelation of the first windowing function,

computing the quotient of the autocorrelation of the new first signal and the autocorrelation of the first windowing function to produce an estimate of the autocorrelation function of the first signal in the absence of the effects of the obscuring processes,

noting the time intervals during which the second obscuring process is occurring,

constructing from the second set of time intervals a second windowing function which makes a transition from one towards zero as the second process obscures the second signal and a transition towards one as the second process terminates,

computing the product of the second corrupted signal and the second windowing function to create a new second signal,

computing the crosscorrelation of the two new signals,

computing the crosscorrelation of the two windowing functions,

computing the quotient of the crosscorrelation of the new signals and the crosscorrelation of the windowing functions to produce an estimate of the crosscorrelation function of the two signals in the absence of the effects of the obscuring processes, and

directly computing an estimate of the impulse response function in the absence of the effects of the obscuring processes from the estimates of the crosscorrelation and the autocorrelation functions.

13. The method of claim 12 further comprising computing the Fourier transform of the estimate of the impulse response function to generate an estimate of the transfer function between the first and second signals in the absence of the effects of the obscuring processes.

14. Method for characterizing the fluctuations between a first signal representing a first underlying biological process, said first signal being intermittently obscured by a first process resulting in a first corrupted signal, and a second signal representing a second underlying biological process, said second signal being intermittently obscured by a second process (which may or may not be the same as the first obscuring process) resulting in a second corrupted signal in which each of the first and second signals are relatively independent of each of the first and second obscuring processes comprising:

noting the time intervals during which the first obscuring process is occurring,

constructing from the intervals a first windowing function which makes a transition from one towards zero as the first process obscures the first signal and a transition towards one as the first process terminates,

computing the product of the first corrupted signal and the first windowing function to create a new first signal,

computing the autocorrelation of the new first signal,

computing the autocorrelation of the first windowing function,

computing the quotient of the autocorrelation of the new first signal and the autocorrelation of the first windowing function to produce an estimate of the autocorrelation function of the first signal in the absence of the effects of the obscuring processes,

noting the time intervals during which the second obscuring process is occuring,

constructing from the second set of time intervals a second windowing function which makes a transition from one towards zero as the second process obscures the second signal and a transition towards one as the second process terminates,

computing the product of the second corrupted signal with the second windowing function to create a new second signal,

computing the crosscorrelation of the two new signals,

computing the crosscorrelation of the two windowing functions,

computing the quotient of the crosscorrelation of the new signals and the crosscorrelation of the windowing functions to generate an estimate of the crosscorrelation function of the two signals in the absence of the effects of the obscuring processes,

computing an estimate of the cross-spectrum of the first and second signals in the absence of the effects of the obscuring processes from the estimate of the crosscorrelation function, and

directly computing an estimate of the transfer function of the first and second signals in the absence of the effects of the obscuring processes from the estimates of the crosscorrelation function and the autocorrelation function.

19. Method for characterizing the fluctuations of a signal representing an underlying biological process, said signal being intermittently obscured by a relatively independent process resulting in a corrupted signal comprising:

noting the time intervals during which the obscuring process is occurring,

constructing from the time intervals a windowing function which makes a transition from one towards zero as the process obscures the signal and a transition towards one as the process terminates,

computing the product of the corrupted signal and the windowing function to create a new signal,

computing the autocorrelation of the new signal,

computing the autocorrelation of the windowing function,

computing the quotient of the autocorrelation of the new signal and the autocorrelation of the windowing function, and

computing an estimate of the power spectrum of the signal in the absence of the effect of the obscuring process from the quotient.