Method and apparatus for continuously measuring volumetric flow

• US 5,146,414 A
• Filed: 04/18/1990
• Issued: 09/08/1992
• Est. Priority Date: 04/18/1990
• Status: Expired due to Term

• Current Assignee: Edwards Lifesciences Corporation
• Original Assignee: Interflo Medical, Inc.
• Inventors: Quinn, Michael, McKown, Russel, Yelderman, Mark
• Primary Examiner(s): Shaw, Dale M.
• Assistant Examiner(s): Melnick, S. A.
• Application Number:
  US07/510,897
• Time in Patent Office: 874 Days
• Field of Search: 364/554, 364/510, 364/413.03, 364/485, 364/484, 128/660.01, 128/660.08, 128/691, 128/713, 73/861.06, 73/861.05
• US Class Current: 702/49
• CPC Class Codes: A61B 5/026 Measuring blood flow A61B3/1233, A61B3/1241 take precedence
• CPC Class Codes: A61B 5/7257 using Fourier transforms
• CPC Class Codes: G01F 1/712 using auto-correlation or cross-correlation detection means

Abstract

Method and apparatus for adapting the volumetric flow rate measurement system taught by Yelderman in U.S. Pat. No. 4,507,974 to clinical environments by reducing the effects of physiological noise. Signal processing techniques are used to characterize the background noise power spectrum of the system under test so that the effect of noise aberrations on the measured data may be eliminated and so that the predominant periodicities of the background noise spectra may be avoided. The invention further cross-correlates the input data of the system with the corresponding ouput data of the system to determine weighting values for each input frequency so that data collected at noisy frequencies is discounted. Low frequency noise or drift is also removed from the output signal by fitting the average of the output data to a quadratic function which is then subtracted from the original output signal, point by point. Finally, the processed data is fit to a lagged normal normal distribution curve in the frequency domain to solve for the parameters necessary to mathematically model the transfer function of the system. Given the input to the system as well as the frequency domain transfer function and noise data of the system, the volumetric flow of the system then may be determined for virtually any input even in very noisy environments without knowledge of the volume of the system.

Claims

1. A method of directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising the steps of:

applying to said fluid, at said system entry point, a quantity of indicator under control of an excitation signal having a plurality of frequencies;

sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

cross-correlating said excitation signal with said response signal to obtain a cross-correlation function signal representing cross-correlation data;

selecting a distribution function model which mathematically models the distribution of said indicator in said system;

comparing said cross-correlation data to said distribution function model;

adjusting said distribution function model to fit said cross-correlation data; and

calculating a signal representative of volumetric flow of said fluid from said quantity of said indicator applied under control of said excitation signal and from said adjusted distribution function model.

2. The method according to claim 1, wherein said fluid is human blood.

3. The method according to claim 2, wherein said indicator is one of heat, cold, a dye, and a radioactive isotope added to said blood.

4. The method according to claim 1, wherein said excitation signal is a stochastic signal.

5. The method according to claim 4, wherein said stochastic signal is a pseudorandom maximum length binary sequence.

6. The method according to claim 1, wherein said excitation signal has at least three frequencies.

7. The method according to claim 6, wherein said excitation signal has fifteen frequencies.

8. The method according to claim 1, comprising the further steps of:

determining substantially noise-free and substantially noisy frequencies of said excitation signal;

calculating from said cross-correlation data weighting values for each frequency of said excitation signal, whereby said weighting values have values approaching one for said substantially noise-free frequencies of said excitation signal and values approaching zero for said substantially noisy frequencies of said excitation signal;

weighting said cross-correlation data with said weighting values so as to substantially eliminate the effect of received data at frequencies containing a predetermined amount of noise; and

outputting said weighted cross-correlation function signal for use in said comparing step.

9. The method according to claim 8, wherein said weighting values calculating step comprises the steps of:

comparing input data at each frequency of said excitation signal with output data at each corresponding frequency component of said response signal;

selecting a coherence factor for each said frequency of said excitation signal, said coherence factor having a value approaching one when said output data is substantially noise-free at a particular frequency of said excitation signal, and a value approaching zero when said output data is substantially noisy at said particular frequency of said excitation signal; and

determining said weighting values for each frequency of said excitation signal in accordance with said coherence factors.

10. The method according to claim 9, wherein said weighting values determining step includes the step of calculating said weighting values as the nth power of the values of said coherence factors, where n is at least 2.

11. The method according to claim 1, wherein said comparing step comprises the step of comparing said cross-correlation data to amplitude and phase functions of a lagged normal distribution function which mathematically models the distribution of said indicator in said system, and said adjusting step comprises the step of fitting said lagged normal distribution function to excitation signal - response function transfer function data, S_xy /S_xx, by adjusting values of said lagged normal distribution function to fit said transfer function data.

12. The method according to claim 1, comprising the further steps of:

obtaining a background noise spectra of the system;

determining the frequencies of predominant noise amplitudes of the background noise spectra; and

selecting said frequencies of said excitation signal so as to differ from said frequencies of said predominant noise amplitudes.

13. The method according to claim 12, wherein the step of determining the frequencies of predominant noise amplitudes of the background noise spectra comprises the step of taking a Fourier transform of the background noise spectra over multiple periods of the excitation signal.

14. The method according to claim 13, wherein said frequencies selecting step comprises the step of determining a clock time of the frequencies of said excitation signal such that said frequencies of predominant noise amplitudes of the background noise spectra are intermediate said frequencies of said excitation signal.

15. The method according to claim 1, comprising the further steps of:

(a) determining the average power of response signals resulting from the input of excitation signals comprising three adjacent pseudorandom noise sequences;

(b) fitting a quadratic curve to data points corresponding to the average power of said response signals resulting from the input of said three adjacent pseudorandom noise sequences;

(c) subtracting a portion of the fitted quadratic curve associated with the response signal for an intermediate pseudorandom noise sequence of said three adjacent pseudorandom noise sequences, data point by data point, from the response signal for said intermediate pseudorandom noise sequence so as to obtain drift adjusted response data;

(d) selecting three different adjacent pseudorandom noise sequences; and

(e) repeating steps (a)-(d) for each of said selected three adjacent pseudorandom noise sequences so as to obtain a drift adjusted response signal.

16. The method according to claim 1, comprising the further steps of:

calculating an RMS value of the response signal for an excitation signal applied to said fluid;

calculating a mean, RMS_mean, and a standard deviation, RMS.sub.σ

, of said RMS value;

determining a threshold, RMS_th, according to the equation RMS_th =RMS_mean +K*RMS.sub.σ

, where K is a predetermined constant; and

removing from said response signal data having an RMS value greater than RMS_th, thereby removing data from said response signal which was caused by an aberrant noise event of said system.

17. The method according to claim 1, wherein said excitation signal has a number of time and frequency intervals and comprises a number of data bits, and said calculating step comprises the step of generating said volumetric flow signal according to the expression:

##EQU52## and P is the rate of indicator infusion per unit time,

C is the specific density of the indicator volume in said fluid,

N is the number of time and frequency intervals in said excitation signal,

T_c is the length of each data bit of said excitation signal,

φ

_xy (s) is a cross-correlation curve representing the cross-correlation function signal over time, and

s is an integration variable.

18. The method according to claim 17, wherein said calculating step comprises the further step of calculating an area under the φ

_xy (s) curve by performing the steps of;

advancing the φ

_xy (s) curve by one T_c period in the time domain;

determining a minimum value in the first two T_c periods from a time origin of said φ

_xy (s) curve as a baseline of said φ

_xy (s) curve;

subtracting said baseline from said φ

_xy (s) curve to provide a curve φ

_xy (s)'"'"'; and

calculating the area under the φ

_xy (s) curve in accordance with the following expression;

##EQU53## where s₂ =τ

_peak +2*σ

+2*T_c, σ

is the standard deviation of the φ

_xy (s)'"'"' curve, τ

_peak is the time of the peak of the φ

_xy (s)'"'"' curve, and τ

is an exponential decay constant of said φ

_xy (s)'"'"' curve.

19. The method according to claim 1, wherein said calculating step comprises the step of generating said volumetric flow signal according to the expression:

##EQU54## where ω

is each respective frequency of said excitation signal,

k₃ =2P/C, P being the rate of indicator infusion per unit time, and C being the specific density of indicator volume in said fluid, where C=4180·

2·

c·

ρ

·

γ

/60, and where c is the specific heat for said fluid, ρ

is the density of said fluid, and γ

is the specific gravity of said fluid,

H(ω

) is the transfer function of said system,

S_xx (ω

) is the spectral density of said excitation signal, and

S_xy (ω

) is the cross-spectral density of said excitation and response signals.

20. The method according to claim 19, wherein said transfer function H(ω

) is represented as a lagged normal distribution equation for said indicator in said fluid.

21. The method according to claim 20, wherein said transfer function is expressed as:

##EQU55## where;

μ

is a time delay of a Gaussian portion of a lagged normal curve corresponding to said lagged normal distribution equation,

σ

is the standard deviation of said lagged normal curve, and

τ

is an exponential decay parameter of said lagged normal curve.

22. The method according to claim 19, wherein said excitation signal has a number of time and frequency intervals and comprises a number of data bits, and said calculating step further comprises the step of minimizing a system identification cost function using a real scalar C defined as:

##EQU56## where;

N corresponds to the number of frequencies of said excitation signal,

T_c is the length of each data bit of said excitation signal, and

SNR(n) is said weighting value for each frequency n of said excitation signal.

23. The method according to claim 22, wherein said step of minimizing a system identification cost function includes the step of deriving a value for SNR(n) at each frequency n of said excitation signal from a coherence function according to the following equation:

##EQU57## where;

P_yy is the power spectral density of the system output estimated using a discrete Fourier transform with windows of length 2NT_c seconds,

Ω

=π

/T_c, and

γ

is a coherence function estimate of said excitation and response signals.

24. A method of directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising the steps of:

applying to said fluid, at said system entry point, a quantity of indicator under control of an excitation signal having a plurality of frequencies;

sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

cross-correlating said excitation signal with said response signal to obtain a cross-correlation function signal representing cross-correlation data;

determining substantially noise-free and substantially noisy frequencies of said excitation signal;

calculating from said cross-correlation data weighting values for each frequency of said excitation signal, whereby said weighting values have values approaching one for said substantially noise-free frequencies of said excitation signal and values approaching zero for said substantially noisy frequencies of said excitation signal;

weighting said cross-correlation data with said weighting values so as to substantially eliminate the effect of received data at frequencies containing a predetermined amount of noise; and

calculating a signal representative of volumetric flow of said fluid from said weighted cross-correlation data and from said quantity of said indicator applied under control of said excitation signal.

25. The method according to claim 24, wherein said weighting values calculating step comprises the steps of:

comparing input data at each frequency of said excitation signal with output data at each corresponding frequency component of said response signal;

selecting a coherence factor for each said frequency of said excitation signal, said coherence factor having a value approaching one when said output data is substantially noise-free at a particular frequency of said excitation signal, and a value approaching zero when said output data is substantially noisy at said particular frequency of said excitation signal; and

determining said weighting values for each frequency of said excitation signal in accordance with said coherence factors.

26. The method according to claim 25, wherein said weighting values determining step includes the step of calculating said weighting values as the nth power of the values of said coherence factors, where n is at least 2.

27. A method of directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising the steps of:

obtaining a background noise spectra of the system;

determining the frequencies of predominant noise amplitudes of the background noise spectra;

selecting frequencies of an excitation signal so as to differ from said frequencies of said predominant noise amplitudes;

applying to said fluid, at said system entry point, a quantity of indicator under control of said excitation signal having said selected frequencies;

sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

cross-correlating said excitation signal with said response signal to obtain a cross-correlation function signal representing cross-correlation data; and

calculating a signal representative of volumetric flow of said fluid from said cross-correlation data and from said quantity of said indicator applied under control of said excitation signal.

28. The method according to claim 27, wherein the step of determining the frequencies of predominant noise amplitudes of the background noise spectra comprises the step of taking a Fourier transform of the background noise spectra over multiple periods of the excitation signal.

29. The method according to claim 28, wherein said frequencies selecting step comprises the step of determining a clock time of the frequencies of said excitation signal such that said frequencies of predominant noise amplitudes of the background noise spectra are intermediate said frequencies of said excitation signal.

30. A method of directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising the steps of:

applying to said fluid, at said system entry point, a quantity of indicator under control of an excitation signal having a plurality of frequencies;

sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

cross-correlating said excitation signal with said response signal to obtain a cross-correlation function signal representing cross-correlation data;

obtaining a background noise spectra of the system;

determining the frequencies of predominant noise amplitudes of the background noise spectra;

selecting said frequencies of said excitation signal so as to differ from said frequencies of said predominant noise amplitudes; and

calculating a signal representative of volumetric flow of said fluid from said cross-correlation data and from said quantity of said indicator applied under control of said excitation signal at said selected frequencies.

31. The method according to claim 30, wherein the step of determining the frequencies of predominant noise amplitudes of the background noise spectra comprises the step of taking a Fourier transform of the background noise spectra over multiple periods of the excitation signal.

32. The method according to claim 31, wherein said frequencies selecting step comprises the step of determining a clock time of the frequencies of said excitation signal such that said frequencies of predominant noise amplitudes of the background noise spectra are intermediate said frequencies of said excitation signal.

33. A method of directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising the steps of:

(a) applying to said fluid, at said system entry point, a quantity of indicator under control of excitation signals comprising three adjacent pseudorandom noise sequences;

(b) sensing a time-dependent response of said fluid to said excitation signals at said system exit point to obtain response signals having a plurality of data points;

(c) determining the average power of response signals resulting from the input of said three adjacent pseudorandom noise sequences;

(d) fitting a quadratic curve to data points corresponding to the average power of said response signals resulting from the input of said three adjacent pseudorandom noise sequences;

(e) subtracting a portion of the fitted quadratic curve associated with the response signal for an intermediate pseudorandom noise sequence of said three adjacent pseudorandom noise sequences, data point by data point, from the response signal for said intermediate pseudorandom noise sequence so as to obtain drift adjusted response data;

(f) selecting three different adjacent pseudorandom noise sequences;

(g) repeating steps (c)-(f) for each of said selected pseudorandom noise sequences so as to obtain a drift adjusted response signal;

(h) cross-correlating said excitation signal with said drift adjusted response signal to obtain a cross-correlation function signal representing cross-correlation data; and

(i) calculating a signal representative of volumetric flow of said fluid from said cross-correlation data and from said quantity of said indicator applied under control of said excitation signal.

34. A method of directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising the steps of:

applying to said fluid, at said system entry point, a quantity of indicator under control of an excitation signal having a plurality of frequencies;

sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

removing from said response signal data representing an aberrant noise event so as to obtain a noise event edited response signal, said aberrant noise event data including a spike in said response signal having an abnormally large amplitude over a short time interval with respect to the remainder of said response signal;

cross-correlation said excitation signal with said noise event edited response signal to obtain a cross-correlation function signal representing cross-correlation data; and

calculating a signal representative of volumetric flow of said liquid from said cross-correlation data and from said quantity of said indicator applied under control of said excitation signal.

35. The method according to claim 34, wherein said step of removing aberrant noise event data includes the steps of:

calculating an RMS value of the response signal for an excitation signal applied to said fluid;

calculating a mean, RMS_mean, and a standard deviation, RMS.sub.σ

, of said RMS value;

determining a threshold, RMS_th, according to the equation RMS_th =RMS_mean +K*RMS.sub.σ

, where K is a predetermined constant; and

removing from said response signal data having an RMS value greater than RMS_th.

36. A method of directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising the steps of:

applying to said fluid, at said system entry point, a quantity of indicator under control of an excitation signal having a plurality of frequencies;

sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

cross-correlating said excitation signal with said response signal to obtain a cross-correlation function signal representing cross-correlation data; and

calculating a signal representative of volumetric flow of said fluid from said quantity of said indicator applied under control of said excitation signal and from a transfer function of said system, wherein said calculating step comprises the step of generating said volumetric flow signal according to the expression;

##EQU58## where;

ω

is each respective frequency of said excitation signal,

k₃ =2P/C, P being the rate of indicator infusion per unit time, and C being the specific density of indicator volume in said fluid, where C=4180·

2·

c·

ρ

·

γ

/60, and where c is the specific heat for said fluid, ρ

is the density of said fluid, and γ

is the specific gravity of said fluid,

H(ω

) is the transfer function of said system,

S_xx (ω

) is the spectral density of said excitation signal, and

S_xy (ω

) is the cross-spectral density of said excitation and response signals.

37. The method according to claim 36, wherein said transfer function H(ω

) is represented as a lagged normal distribution equation for said indicator in said fluid.

38. The method according to claim 37, wherein said transfer function is expressed as:

##EQU59## where;

μ

is a time delay of a Gaussian portion of a lagged normal curve corresponding to said lagged normal distribution equation,

σ

is the standard deviation of said lagged normal curve, and

τ

is an exponential decay parameter of said lagged normal curve.

39. The method according to claim 36, wherein said excitation signal has a number of time and frequency intervals and comprises a number of data bits, and said calculating step includes the further step of minimizing s system identification cost function using a real scalar C defined as:

##EQU60## where;

N corresponds to the number of frequencies of said excitation signal,

T_c is the length of each data bit of said excitation signal, and

SNR(n) is said weighting value for each frequency n of said excitation signal.

40. The method according to claim 39, wherein said step of minimizing a system identification cost function includes the step of deriving a value for SNR(n) at each frequency n of said excitation signal from a coherence function according to the following equation:

##EQU61## where;

P_yy is the power spectral density of the system output estimated using a discrete Fourier transform with windows of length 2NT_c seconds,

Ω

=π

/T_c, and

γ

is a coherence function estimate of said excitation and response signals.

41. A method of directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising the steps of:

(a) determining a distribution function model which mathematically model the distribution of an indicator in said system;

(b) applying to said fluid, at said system entry point, a quantity of said indicator under control of an excitation signal having a plurality of frequencies;

(c) sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

(d) cross-correlating said excitation signal with said response signal to obtain a cross-correlation function signal representing cross-correlation data;

(e) comparing said cross-correlation data to said distribution function model;

(f) adjusting said distribution function model to fit said cross-correlation data;

(g) calculating a signal representative of volumetric flow of said fluid from said quantity of said indicator applied under control of said excitation signal and from said adjusted distribution model;

(h) repeating said steps (b) through (g) to produce a plurality of said volumetric flow signals; and

(i) averaging said plurality of volumetric flow signals to produce an average volumetric flow signal.

42. An apparatus for directly measuring volumetric flow of a fluid between a system entry point and a system exit pint, comprising:

means for applying to said fluid, at said system entry point, a quantity of indicator under control of an excitation signal having a plurality of frequencies;

means for sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

means for cross-correlating said excitation signal with said response signal to obtain a cross-correlation function signal representing cross-correlation data;

means for comparing said cross-correlation data to a distribution function model which mathematically models the distribution of said indicator in said system;

means for adjusting said distribution function model to fit said cross-correlation data; and

means for calculating a signal representative of volumetric flow of said fluid from said quantity of said indicator applied under control of said excitation signal and from said adjusted distribution function model.

43. The apparatus of claim 42, wherein said fluid is human blood.

44. The apparatus of claim 43, wherein said indicator is one of heat, cold, a dye, and a radioactive isotope added to said blood.

45. The apparatus of claim 42, wherein said excitation signal is a stochastic signal.

46. The apparatus of claim 45, wherein said stochastic signal is a pseudorandom maximum length binary sequence.

47. The apparatus of claim 42, wherein said excitation signal has at least three frequencies.

48. The apparatus of claim 47, wherein said excitation signal has fifteen frequencies.

49. The apparatus of claim 42, further comprising:

means for determining substantially noise-free and substantially noisy frequencies of said excitation signal;

means for calculating from said cross-correlation data weighting values for each frequency of said excitation signal, whereby said weighting values have values approaching one for said substantially noise-free frequencies of said excitation signal and values approaching zero for said substantially noisy frequencies of said excitation signal;

means for weighting said cross-correlation data with said weighting values so as to substantially eliminate the effect of received data at frequencies containing a predetermined amount of noise; and

means for outputting said weighted cross-correlation function signal to said comparing means.

50. The apparatus of claim 49, wherein said weighting values calculating means comprises weighting value processing means which performs the steps of:

comparing input data at each frequency of said excitation signal with output data at each corresponding frequency component of said response signal;

selecting a coherence factor for each said frequency of said excitation signal, said coherence factor having a value approaching one when said output data is substantially noise-free at a particular frequency of said excitation signal, and a value approaching zero when said output data is substantially noisy at said particular frequency of said excitation signal; and

determining said weighting values for each frequency of said excitation signal in accordance with said coherence factors.

51. The apparatus of claim 50, wherein said weighting values calculating means calculates said weighting values as the nth power of the values of said coherence factors as said weighting values, where n is at least 2.

52. The apparatus of claim 42, wherein said comparing means compares said cross-correlation data to amplitude and phase functions of a lagged normal distribution function which mathematically models the distribution of said indicator in said system, and said adjusting means fits said lagged normal distribution function to excitation signal - response signal transfer function data, S_xy /S_xx, by adjusting values of said lagged normal distribution function to fit said transfer function data.

53. The apparatus of claim 42, further comprising setup means for obtaining a background noise spectra of the system, determining the frequencies of predominant noise amplitudes of the background noise spectra, and selecting said frequencies of said excitation signal so as to differ from said frequencies of said predominant noise amplitudes.

54. The apparatus of claim 53, wherein said setup means comprises means for taking a Fourier transform of the background noise spectra over multiple periods of the excitation signal for determining the frequencies of predominant noise amplitudes of the background noise spectra.

55. The apparatus of claim 54, wherein said setup means comprises means for determining a clock time of the frequencies of said excitation signal such that said frequencies of predominant noise amplitudes of the background noise spectra are intermediate said frequencies of said excitation signal.

56. The apparatus of claim 42, wherein said excitation signal comprises three adjacent pseudorandom noise sequences and said apparatus further comprises drift removal means including processing means for removing low frequency drift of said response signal by performing the steps of:

(a) determining the average power of response signals resulting from the input of said three adjacent pseudorandom noise sequences;

(b) fitting a quadratic curve to data points corresponding to the average power of said response signals resulting from the input of said three adjacent pseudorandom noise sequences;

(c) subtracting a portion of the fitted quadratic curve associated with the response signal for an intermediate pseudorandom noise sequence of said three adjacent pseudorandom noise sequences, data point by data point, from the response signal for said intermediate pseudorandom noise sequence so as to obtain drift adjusted response data;

(d) selecting three different adjacent pseudorandom noise sequences; and

(e) repeating steps (a)-(d) for each of said selected three adjacent pseudorandom noise sequences so as to obtain a drift adjusted response signal.

57. The apparatus of claim 42, further comprising noise event editing means including processing means for performing the steps of:

calculating an RMS value of the response signal for an excitation signal applied to said fluid;

calculating a mean, RMS_mean, and a standard deviation, RMS.sub.σ

, of said RMS value;

determining a threshold, RMS_th, according to the equation RMS_th =RMS_mean +K*RMS.sub.σ

, where K is a predetermined constant; and

removing from said response signal data having an RMS value greater than RMS_th, thereby removing data from said response signal which was caused by an aberrant noise event of said system.

58. The apparatus of claim 42, wherein said excitation signal has a number of time and frequency intervals and comprises a number of data bits, and said calculating means comprises processing means for generating said volumetric flow signal according to the expression:

##EQU62## and P is the rate of indicator infusion per unit time,

C is the specific density of the indicator volume in said fluid,

N is the number of time and frequency intervals in said excitation signal,

T_c is the length of each data bit of said excitation signal,

φ

_xy (s) is a cross-correlation curve representing the cross-correlation function signal over time, and

s is an integration variable.

59. The apparatus of claim 58, wherein said processing means further calculates an area under the φ

_xy (s) curve by performing the steps of;

advancing the φ

_xy (s) curve by one T_c period in the time domain;

determining a minimum value in the first two T_c periods from a time origin of said φ

_xy (s) curve as a baseline of said φ

_xy (s) curve;

subtracting said baseline from said φ

_xy (s) curve to provide a curve φ

_xy (s)'"'"'; and

calculating the area under the φ

_xy (s) curve in accordance with the following expression;

##EQU63## where s₂ =τ

_peak +2*σ

+2*T_c, σ

is the standard deviation of the φ

_xy (s)'"'"' curve, τ

_peak is the time of the peak of the φ

_xy (s)'"'"' curve, and τ

is an exponential decay constant of said φ

_xy (s)'"'"' curve.

60. The apparatus of claim 42, wherein said calculating means comprises means for generating said volumetric flow signal according to the expression:

##EQU64## where ω

is each respective frequency of said excitation signal,

k₃ =2P/C, P being the rate of indicator infusion per unit time, and C being the specific density of indicator volume in said fluid, where C=4180·

2·

c·

ρ

·

γ

/60, and where c is the specific heat for said fluid, ρ

is the density of said fluid, and γ

is the specific gravity of said fluid,

H(ω

) is the transfer function of said system,

S_xx (ω

) is the spectral density of said excitation signal, and

S_xy (ω

) is the cross-spectral density of said excitation and response signals.

61. The apparatus of claim 60, wherein said transfer function H(ω

) is represented as a lagged normal distribution equation for said indicator in said fluid.

62. The apparatus of claim 61, wherein said transfer function is expressed by said generating means as:

##EQU65## where;

μ

is a time delay of a Gaussian portion of a lagged normal curve corresponding to said lagged normal distribution equation,

σ

is the standard deviation of said lagged normal curve, and

τ

is an exponential decay parameter of said lagged normal curve.

63. The apparatus of claim 60, wherein said excitation signal has a number of time and frequency intervals and comprises a number of data bits, and said means for generating said volumetric flow signal comprises means for minimizing a system identification cost function using a real scalar C defined as:

##EQU66## where;

N corresponds to the number of frequencies of said excitation signal,

T_c is the length of each data bit of said excitation signal, and

SNR(n) is said weighting value for each frequency n of said excitation signal.

64. The apparatus of claim 63, wherein said minimizing means comprises means for deriving a value for SNR(n) for each frequency n of said excitation signal from a coherence function according to the following equation:

##EQU67## where;

P_yy is the power spectral density of the system output estimated using a discrete Fourier transform with windows of length 2NT_c seconds,

Ω

=π

/T_c, and

γ

is a coherence function estimate of said excitation and response signals.

65. An apparatus for directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising:

means for applying to said fluid, at said system entry point, a quantity of indicator under control of an excitation signal having a plurality of frequencies;

means for sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

means for cross-correlating said excitation signal with said response signal to obtain a cross-correlation function signal representing cross-correlation data;

means for determining substantially noise-free and substantially noisy frequencies of said excitation signal;

means for calculating from said cross-correlation data weighting values for each frequency of said excitation signal, whereby said weighting values have values approaching one for said substantially noise-free frequencies of said excitation signal and values approaching zero for said substantially noisy frequencies of said excitation signal;

means for weighting said cross-correlation data with said weighting values so as to substantially eliminate the effect of received data at frequencies containing a predetermined amount of noise; and

means for calculating a signal representative of volumetric flow of said fluid from said weighted cross-correlation data and from said quantity of said indicator applied under control of said excitation signal.

66. The apparatus of claim 65, wherein said weighting values calculating means comprises:

means for comparing input data at each frequency of said excitation signal with output data at each corresponding frequency component of said response signal;

means for selecting a coherence factor for each said frequency of said excitation signal, said coherence factor having a value approaching one when said output data is substantially noise-free at a particular frequency of said excitation signal, and a value approaching zero when said output data is substantially noisy at said particular frequency of said excitation signal; and

means for determining said weighting values for each frequency of said excitation signal in accordance with said coherence factors.

67. The apparatus of claim 66, wherein said weighting values determining means includes means for calculating said weighting values as the nth power of the values of said coherence factors, where n is at least 2.

68. An apparatus for directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising:

setup means for obtaining a background noise spectra of the system, determining the frequencies of predominant noise amplitudes of the background noise spectra, and selecting frequencies of an excitation signal so as to differ from said frequencies of said predominant noise amplitudes;

means for applying to said fluid, at said system entry point, a quantity of indicator under control of said excitation signal having said selected frequencies;

means for sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

means for cross-correlating said excitation signal with said response signal to obtain a cross-correlation function signal representing cross-correlation data; and

means for calculating a signal representative of volumetric flow of said fluid from said cross-correlation data and from said quantity of said indicator applied under control of said excitation signal.

69. The apparatus of claim 68, wherein said setup means comprises means for taking a Fourier transform of the background noise spectra over multiple periods of the excitation signal for determining the frequencies of predominant noise amplitudes of the background noise spectra.

70. The apparatus of claim 69, wherein said setup means comprises means for determining a clock time of the frequencies of said excitation signal such that said frequencies of predominant noise amplitudes of the background noise spectra are intermediate said frequencies of said excitation signal.

71. An apparatus for directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising:

means for applying to said fluid, at said system entry point, a quantity of indicator under control of excitation signals comprising three adjacent pseudorandom noise sequences;

means for sensing a time-dependent response of said fluid to said excitation signals at said system exit point to obtain response signals having a plurality of data points;

drift removal means including processing means for removing low frequency drift of said response signal by performing the steps of;

(a) determining the average power of response signals resulting from the input of said three adjacent pseudorandom noise sequences;

(b) fitting a quadratic curve to points corresponding to the average power of said response signals resulting from the input of said three adjacent pseudorandom noise sequences;

(c) subtracting a portion of the fitted quadratic curve associated with the response signal for an intermediate pseudorandom noise sequence of said three adjacent pseudorandom noise sequences, data point by data point, from the response signal for said intermediate pseudorandom noise sequence so as to obtain drift adjusted response data;

(d) selecting three different adjacent pseudorandom noise sequences; and

(e) repeating steps (a)-(d) for each of said selected pseudorandom noise sequences so as to obtain a drift adjusted response signal;

means for cross-correlating said excitation signal with said drift adjusted response signal to obtain a cross-correlation function signal representing cross-correlation data; and

means for calculating a signal representative of volumetric flow of said fluid from said cross-correlation data and from said quantity of said indicator applied under control of said excitation signal.

72. An apparatus for directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising:

means for applying to said fluid, at said system entry point, a quantity of indicator under control of an excitation signal having a plurality of frequencies;

means for sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

means for removing from said response signal data representing an aberrant noise event so as to obtain a noise event edited response signal, said aberrant noise event data including a spike in said response signal having an abnormally large amplitude over a short time interval with respect to the remainder of said response signal;

means for cross-correlating said excitation signal with said noise event edited response signal to obtain a cross-correlation function signal representing cross-correlation data; and

means for calculating a signal representative of volumetric flow of said fluid from said cross-correlation data and from said quantity of said indicator applied under control of said excitation signal.

73. The apparatus of claim 72, wherein said aberrant noise event removing means comprises:

processing means for performing the steps of;

calculating an RMS value of the response signal for an excitation signal applied to said fluid,

calculating a mean, RMS_mean, and a standard deviation, RMS.sub.σ

, of said RMS value, and

determining a threshold, RMS_th, according to the equation RMS_th =RMS_mean +K*RMS.sub.σ

, where K is a predetermined constant; and

means for removing from said response signal data having an RMS value greater than RMS_th.

74. An apparatus for directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising:

means for applying to said fluid, at said system entry point, a quantity of indicator under control of an excitation signal having a plurality of frequencies;

means for sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

means for cross-correlating said excitation signal with said response signal to obtain a cross-correlation function signal representing cross-correlation data; and

means for calculating a signal representative of volumetric flow of said fluid from said quantity of said indicator applied under control of said excitation signal and from a transfer function of said system, said calculating means comprising processing means for generating said volumetric flow signal according to the expression;

##EQU68## where;

ω

is each respective frequency of said excitation signal,

k₃ =2P/C, P being the rate of indicator infusion per unit time, and C being the specific density of indicator volume in said fluid, where C=4180·

2·

c·

ρ

·

γ

/60, and where c is the specific heat for said fluid, ρ

is the density of said fluid, and γ

is the specific gravity of said fluid,

H(ω

) is the transfer function of said system,

S_xx (ω

) is the spectral density of said excitation signal, and

S_xy (ω

) is the cross-spectral density of said excitation and response signals.

75. The apparatus of claim 74, wherein said transfer function H(ω

) is represented as a lagged normal distribution equation for said indicator in said fluid.

76. The apparatus of claim 75, wherein said transfer function is expressed by said processing means as:

##EQU69## where;

μ

is a time delay of a Gaussian portion of a lagged normal curve corresponding to said lagged normal distribution equation,

σ

is the standard deviation of said lagged normal curve, and

τ

is an exponential decay parameter of said lagged normal curve.

77. The apparatus of claim 74, wherein said excitation signal has a number of time and frequency intervals and comprises a number of data bits, and said processing means comprises means for minimizing a system identification cost function using a real scalar C defined as:

##EQU70## where;

N corresponds to the number of frequencies of said excitation signal,

T_c is the length of each data bit of said excitation signal, and

SNR(n) is said weighting value for each frequency n of said excitation signal.

78. The apparatus of claim 77, wherein said minimizing means comprises means for deriving a value for SNR(n) for each frequency n of said excitation signal from a coherence function according to the following equation:

##EQU71## where;

P_yy is the power spectral density of the system output estimated using a discrete Fourier transform with windows of length 2NT_c seconds,

Ω

=π

/T_c, and

γ

is a coherence function estimate of said excitation and response signals.

79. An apparatus for directly measuring volumetric flow of a fluid between a system entry point and a system exit point, comprising:

means for repeatedly applying to said fluid, at said system entry point, a quantity of indicator under control of an excitation signal having a plurality of frequencies;

means for sensing a time-dependent response of said fluid to said excitation signal at said system exit point to obtain a response signal having a plurality of data points;

means for cross-correlating said excitation signal with said response signal to obtain a cross-correlation function signal representing cross-correlation data;

means for determining a distribution function model which mathematically models the distribution of an indicator in said system and for comparing said cross-correlation data to said distribution function model;

means for adjusting said distribution function model to fit said cross-correlation data;

means for calculating a signal representative of volumetric flow of said fluid from said quantity of said indicator applied under control of said excitation signal and from said adjusted distribution function model for each application of said quantity of said indicator; and

means for averaging said volumetric flow signals for each application of said quantity of said indicator to produce an average volumetric flow signal.