Blood coagulation analyzer

• US 6,524,861 B1
• Filed: 01/22/1999
• Issued: 02/25/2003
• Est. Priority Date: 01/22/1999
• Status: Expired due to Term

• Current Assignee: Medical Laboratory Automation Incorporated
• Original Assignee: Medical Laboratory Automation Incorporated
• Inventors: Anderson, Eric S.
• Primary Examiner(s): Warden, Jill
• Assistant Examiner(s): Cole, Monique T.
• Application Number:
  US09/235,530
• Time in Patent Office: 1,495 Days
• Field of Search: 436/69, 436/43, 436/47, 436/48, 436/50, 436/55, 422/63, 422/65, 422/67, 422/73, 702/19, 702/21
• US Class Current: 436/69
• CPC Class Codes: G01N 33/86 involving blood coagulating time or factors, or their receptors
• CPC Class Codes: Y10T 436/11 Automated chemical analysis
• CPC Class Codes: Y10T 436/113332 with conveyance of sample along a test line in a container or rack
• CPC Class Codes: Y10T 436/114165 with step of insertion or removal from test line
• CPC Class Codes: Y10T 436/115831 Condition or time responsive
• CPC Class Codes: Y10T 436/12 Condition responsive control

Abstract

One method and apparatus for analyzing clotting characteristics of a blood sample includes the use of a non-linear equation having coefficients capable of being related to the underlying clotting processes. The non-linear equation is curve fit to a waveform of a clotting sample to provide values for the coefficients of the non-linear function. Once the coefficients are obtained, an inference engine may be used to evaluate the non-linear relationship between the coefficients and factor concentrations within the blood sample. Advantageously, the waveform of the actual sample may be detrended to extract a residual oscillatory component to aid in the determination of initial coefficients for simplifying the curve fit operation. The non-linear equation may additionally be used for providing simulated clotting waveforms for testing clot analysis instruments. A residual oscillatory component may advantageously be extracted from the clotting signal. The oscillatory components may be used to provide a high quality clot time indicator for the sample. In addition, the oscillatory component may be used to correct for clot anomalies by correlating the frequency of the oscillatory to various physical characteristics of the blood clot. The oscillatory component may be used to provide initial estimates for values of the coefficients of the logistic. The logistic, because it represents knowledge as to the underlying chemistry of the clotting process, may also be used to provide initial estimates for coefficients of the logistic, and may also be used to enhance the accuracy of existing polynomial template clot analysis techniques. The logistic model may also be used to verify the operation of clot analysis tools.

Claims

1. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing a relationship between initial concentrations of plural blood factors and fibrin polymerization at least in part by one non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multivariate backward correlation; and

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample.

2. The method for determining the physical characteristics of a blood sample according to claim 1, further comprising the step of transforming the values of the plurality of coefficients to provide derived physical parameters of the blood sample.

3. The method for determining the physical characteristics of a blood sample of claim 1 wherein the physical parameters include clot time.

6. The method for determining the physical characteristics of a blood sample according to claim 1 wherein the first signal is an optical density signal.

7. The method for determining the physical characteristics of a blood sample according to claim 1, wherein the first signal is a turbidity signal.

8. The method for determining the physical characteristics of a blood sample of claim 1, wherein the step of curve fitting further comprises the step of:

estimating initial values of the plurality of coefficients of the non-linear function.

15. The method for determining the physical characteristics of a blood sample of claim 1 wherein the physical parameters are selected from a group consisting essentially of initial concentration of blood clotting factors in the blood sample, anti-coagulants, FDP, d-dimer, selected chemistry anti-coagulants and antibodies.

17. The method according to claim 1 wherein the first signal is produced by an optical nephelometer, and the density signal indicates a turbidity of the blood sample.

18. The method according to claim 1 wherein the first signal is produced by a Fiberometer.

19. The method for determining the physical characteristics of a blood sample according to claim 3, wherein the step of transforming further comprises the steps of:

modifying the values of the plurality of coefficients to correct exception conditions; and

translating the modified values of the plurality of coefficients to the derived physical parameters of the blood sample according to a defined operative function.

20. The method for determining the physical characteristics of a blood sample according to claim 19, wherein the means for translating the modified values comprises a multi-variate general linear hypothesis mechanism.

4. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation; and

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

wherein the physical parameters include clot time; and

wherein the clot time substantially correlates with one of the plurality of coefficients representing a centroid location of the non-linear function.

5. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

wherein the physical parameters include clot time; and

wherein the clot time substantially correlates to a centroid location of the second derivative of the non-linear function.

9. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

wherein curve fitting further includes estimating initial values of the plurality of coefficients of the non-linear function; and

wherein estimating initial values determines the values of the plurality of coefficients responsive to characteristics of a low-level oscillatory component of the first signal.

10. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

wherein curve fitting further includes estimating initial values of the plurality of coefficients of the non-linear function; and

wherein estimating initial values determines the values of the plurality of coefficients responsive to characteristics of the first derivative and second derivative of the first signal.

11. The method for determining the physical characteristics of a blood sample of claim 10, wherein the initial values are derived responsive to located vanishing points of the first derivative of the first signal.

12. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

wherein curve fitting further includes estimating initial values of the plurality of coefficients of the non-linear function; and

wherein the non-linear function is a polynomial function, and wherein estimating initial values of the coefficients derives the coefficients responsive to characteristics of a third derivative of the polynomial function.

13. The method for determining the physical characteristics of a blood sample according to claim 12, wherein the step of estimating initial values further comprises the steps of:

searching for vanishing points of the third derivative of the polynomial function to identify bounds of the third derivative of the polynomial function for analysis;

iteratively modifying the bounds of the polynomial to curve fit to the first signal to identify a plurality of potential centroid locations of the polynomial, wherein each of the plurality of identified centroid locations are stored in a table with an associated chi square error of the curve fit;

modifying the error values in the table by applying a penalty value to each of the error values in the table responsive to chaotic characteristics of the first signal; and

selecting one of the centroid locations of the plurality of centroid locations as an initial estimated coefficient responsive to the modified error value associated with the selected centroid location in the table.

14. The method for determining the physical characteristics of a blood sample according to claim 13, wherein the step of searching for vanishing points of the third derivative of the polynomial is controlled responsive to information from the non-linear function.

16. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation; and

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

wherein the non-linear function is $a+\frac{\left(b+\mathrm{ht}\right)}{1+{\left(\frac{t}{c}\right)}^d}+\frac{2<br><br>{\left({\mathrm{eg}}^2<br><br>\left(\frac{t}{f}\right)\right)}^{\left(-2-2<br><br>g\right)}}{{\left(f^2<br><br>\left(1+{\left(\frac{t}{f}\right)}^g\right)\right)}^3}-\frac{\mathrm{eg}<br><br>\left(-1+g\right)<br><br>{\left(\frac{t}{f}\right)}^{\left(-2+g\right)}}{{\left(f_2<br><br>\left(1+{\left(\frac{t}{f}\right)}^g\right)\right)}^2},$

where a, b c, d, e, f and g represent the plurality of coefficients and t represents time.

21. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

transforming the values of the plurality of coefficients to provide derived physical parameters of the blood sample;

wherein transforming further includes;

modifying the values of the plurality of coefficients to correct exception conditions;

translating the modified values of the plurality of coefficients to the derived physical parameters of the blood sample according to a defined operative function; and

wherein translating the modified values is performed using a neural network.

22. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

transforming the values of the plurality of coefficients to provide derived physical parameters of the blood sample;

wherein transforming further includes;

modifying the values of the plurality of coefficients to correct exception conditions;

translating the modified values of the plurality of coefficients to the derived physical parameters of the blood sample according to a defined operative function; and

wherein translating the modified values optimizing is performed using fuzzy logic.

23. A method for determining the physical characteristics of a blood sample, comprising the steps of:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time; and

extracting a low level oscillation signal from the optical density signal, wherein the low level oscillation signal is used to determine the characteristics of the blood sample.

24. The method for determining the physical characteristics of a blood sample according to claim 23, further comprising the step of determining a clot time of the blood sample responsive to the low level oscillation signal.

34. The method for determining the physical characteristics of a blood sample according to claim 23, further comprising the steps of:

providing a mathematical model of the measurement of fibrin polymerization of the blood sample over the given period of time, the mathematical model comprising a plurality of coefficients, wherein the mathematical model is curve fit to the first signal provided by the detector and wherein values of the coefficients may be used to identify physical characteristics of the blood sample; and

determining, responsive to the low level oscillation signal, initial estimates of the coefficients of the mathematical model for the curve fit.

39. The method for determining the physical characteristics of a blood sample according to claim 23 further comprising the step of:

filtering the first signal to reduce a number of the first signals that are transferred to the curve fitter.

25. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

extracting a low level oscillation signal from the first signal, wherein the low level oscillation signal is used to determine the characteristics of the blood sample; and

determining a clot time of the blood sample responsive to the low level oscillation signal;

wherein determining the clot time of the blood sample includes analyzing the frequency of the low level oscillation signal using an analysis method selected from the group consisting of Fourier analysis, fast Fourier transform analysis, wavelet analysis, peak and minimum search analysis and phase locked analysis.

26. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

extracting a low level oscillation signal from the first signal, wherein the low level oscillation signal is used to determine the characteristics of the blood sample;

determining, responsive to the low level oscillation, characteristics of the blood sample; and

correcting characteristic estimations of the blood sample responsive to frequency of the low level oscillation.

27. The method for determining the physical characteristics of a blood sample according to claim 26, wherein the characteristic estimations include fibrinogen estimations.

28. The method for determining the physical characteristics of a blood sample according to claim 26, wherein the step of correcting characteristic estimations includes the step of analyzing the frequency of the low level oscillation signal using an analysis method selected from the group consisting of Fourier analysis, fast Fourier transform analysis, moving fast Fourier transform window analysis, wavelet analysis and peak and minimum search analysis.

29. The method for determining the physical characteristics of a blood sample according to claim 26, wherein the step of correcting characteristic estimations includes the step of inferring relationships between the frequency of the low level oscillation and one or more characteristics to be corrected.

30. The method for determining the physical characteristics of a blood sample according to claim 29, wherein the step of inferring relationships is performed by a trained neural network.

31. The method for determining the physical characteristics of a blood sample according to claim 29, wherein the step of inferring relationships is performed using fuzzy logic.

32. The method for determining the physical characteristics of a blood sample according to claim 29, wherein the step of inferring relationships is performed using one or more multi-variate calibration curves.

33. The method for determining the physical characteristics of a blood sample according to claim 26, wherein the characteristics are selected from a group consisting essentially of fibrin strand mass length ratios, reactant concentrations and abnormalities and kinetic reaction rates of the blood sample.

35. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

extracting a low level oscillation signal from the first signal, wherein the low level oscillation signal is used to determine the characteristics of the blood sample;

providing a mathematical model of the measurement of fibrin polymerization of the blood sample over the given period of time, the mathematical model comprising a plurality of coefficients, wherein the mathematical model is curve fit to the first signal provided by the detector and wherein values of the coefficients may be used to identify physical characteristics of the blood sample; and

determining, responsive to the low level oscillation signal, initial estimates of the coefficients of the mathematical model for the curve fit;

wherein determining the clot time of the initial coefficients includes analyzing the frequency of the low level oscillation signal using an analysis method selected from the group consisting of Fourier analysis, fast Fourier transform analysis, wavelet analysis, peak and minimum search analysis and phase locked loop analysis.

36. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

extracting a low level oscillation signal from the first signal, wherein the low level oscillation signal is used to determine the characteristics of the blood sample;

providing a mathematical model of the measurement of fibrin polymerization of the blood sample over the given period of time, the mathematical model comprising a plurality of coefficients, wherein the mathematical model is curve fit to the first signal provided by the detector and wherein values of the coefficients may be used to identify physical characteristics of the blood sample; and

determining, responsive to the low level oscillation signal, initial estimates of the coefficients of the mathematical model for the curve fit;

wherein the mathematical model is represented by the below equation;

$a+\frac{\left(b+\mathrm{ht}\right)}{1+{\left(\frac{t}{c}\right)}^d}+\frac{2<br><br>{\left({\mathrm{eg}}^2<br><br>\left(\frac{t}{f}\right)\right)}^{\left(-2-2<br><br>g\right)}}{{\left(f^2<br><br>\left(1+{\left(\frac{t}{f}\right)}^g\right)\right)}^3}-\frac{\mathrm{eg}<br><br>\left(-1+g\right)<br><br>{\left(\frac{t}{f}\right)}^{\left(-2+g\right)}}{{\left(f_2<br><br>\left(1+{\left(\frac{t}{f}\right)}^g\right)\right)}^2},$

where a, b c, d, e, f and g represent the plurality of coefficients and t represents time.

37. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample; and

extracting a low level oscillation signal from the first signal, wherein the low level oscillation signal is used to determine the characteristics of the blood sample;

wherein extracting the low level oscillation signal is performed using residuals analysis of the first signal.

38. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample; and

extracting a low level oscillation signal from the first signal, wherein the low level oscillation signal is used to determine the characteristics of the blood sample;

wherein extracting the low level oscillation signal performs a repetitive polynomial detrending of the first signal to retrieve the low level oscillation signal.

40. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

extracting a low level oscillation signal from the first signal, wherein the low level oscillation signal is used to determine the characteristics of the blood sample; and

identifying blood samples having weak clotting characteristics, wherein identifying includes

polynomial detrending which removes external noise from the first signal by adjusting values of a portion the accumulated data signals; and

interpolating which smoothes the interface between said adjusted portion of said accumulated data signals and remaining accumulated data signals.

41. A method for determining the physical characteristics of a blood sample, comprising:

producing a first signal indicative of a measurement of fibrin polymerization of a blood sample over a period of time;

providing an analytical model of fibrin polymerization of the blood sample over a period of time, the model representing at least in part by a non-linear function having a plurality of coefficients which are related to the initial concentration of plural blood factors through multi-variate backward correlation;

curve fitting the model to the first signal to produce an output representing values of the plurality of coefficients causing the model to conform to the first signal, wherein the values of the plurality of coefficients are used to identify physical parameters of the blood sample;

storing, for each manipulation of the bounds of the polynomial model, a difference between the bounded waveform and the first signal, the table also for storing a centroid location of the associated model; and

selecting the bounds of the polynomial model responsive to a non-linear logistic model of the first signal.

42. The method for determining the clotting characteristics of a blood sample of claim 41, further comprising the step of:

selecting one of the centroid locations of one of the plurality of models as a clot time of the blood sample responsive to the corresponding difference signal of the model.

43. The method for determining the clotting characteristics of a blood sample according to claim 41, further comprising the steps of:

storing, in a penalty table, a plurality of penalties, each penalty corresponding to the unit of time of one of the first signals and representing potential chaotic behavior of the first signal at the unit of time; and

applying the plurality of penalties to the differences in the table, wherein one of the plurality of penalties is selected for applying to one of the stored differences responsive to the centroid location of the model.