Radiation apparatus and method for analysis of analytes in sample

• US 6,119,026 A
• Filed: 02/26/1998
• Issued: 09/12/2000
• Est. Priority Date: 12/04/1997
• Status: Expired due to Term

• Current Assignee: Sanofi-Aventis Deutschland GmbH (Sanofi )
• Original Assignee: Hewlett-Packard Company (HP Inc.)
• Inventors: McNulty, Chris, Mauze, Ganapati Ramnath
• Primary Examiner(s): WINAKUR, ERIC FRANK
• Application Number:
  US09/031,369
• Time in Patent Office: 929 Days
• Field of Search: 600/310, 600/316, 600/322, 600/323, 600/326, 600/473, 600/476
• US Class Current: 600/310
• CPC Class Codes: A61B 5/14532 for measuring glucose, e.g. by tissue impedance measurement
• CPC Class Codes: A61B 5/1455 using optical sensors, e.g. spectral photometrical oximeters
• CPC Class Codes: A61B 5/1495 Calibrating or testing of in-vivo probes
• CPC Class Codes: A61B 5/7257 using Fourier transforms
• CPC Class Codes: A61B 5/726 using Wavelet transforms

Abstract

A technique that uses wavelet analysis to analyze the concentration of an analyte in a sample. The technique includes irradiating the sample with electromagnetic radiation and detecting a resulting radiation from the sample to obtain spectral data. The electromagnetic radiation irradiated on the sample results in radiation interaction, such as reflection, scattering, and transmission, from the sample. The radiation interaction results in a resultant electromagnetic radiation from the sample, such as reflected radiation, scattered radiation, or transmitted radiation. The resultant electromagnetic radiation includes signals indicative of the analyte. The spectral data is digitally processed using wavelet analysis to increase proportionally the signal indicative of the analyte in the data. A modeling algorithm is applied to the processed data to determine the quantitative characteristics of the analyte in the sample.

Claims

1. An apparatus for analyzing analyte composition of a sample without immersing a part of the apparatus therein, comprising:

(a) a source of electromagnetic radiation for irradiating the sample to result in radiation interaction;

(b) a detector for detecting electromagnetic radiation resulting from the radiation interaction from the sample to result in analytical data, said detected electromagnetic radiation having a signal indicative of the analyte; and

(c) a processor having a modeling algorithm for applying to said analytical data from step (b) to determine quantitative characteristics of the analyte in the sample, said modeling algorithm having parameters, for determining said quantitative characteristics, obtained by applying wavelet basis functions of resolution levels having amplitude-frequency characteristics resembling amplitude-frequency characteristics of radiation interaction data derived by irradiating calibration samples having known concentration of the analyte.

2. An apparatus according to claim 1 wherein the modeling algorithm applies said wavelet basis functions to calibration data obtained from the detector over a spectrum of frequencies to derive parameters of prediction equations for analyte concentration.

3. An apparatus according to claim 1 wherein the modeling algorithm applies said wavelet basis functions of resolution levels over a selected portion of a spectrum of frequencies of the electromagnetic radiation irradiating the sample for matching characteristic frequencies of the analyte.

4. An apparatus according to claim 1 wherein the radiation source emits light of frequencies that penetrates non-invasively a physiological body surface to irradiate tissue underneath the body surface.

5. An apparatus according to claim 1 wherein the processor derives the parameters from wavelet basis functions of the Daubechies set to correspond to the frequency and amplitude characteristics of the radiation interaction of glucose as the analyte.

6. A method for detecting quantitatively an analyte in a sample, comprising:

(a) irradiating electromagnetic radiation on the sample to result in radiation interaction;

(b) detecting electromagnetic radiation resulting from the radiation interaction from the sample to result in data, said detected electromagnetic radiation having a signal indicative of the analyte; and

(c) applying to the data from step (b) a modeling algorithm to determine quantitative characteristics of the analyte in the sample, said model algorithm having parameters for determining said quantitative characteristics, obtained by applying wavelet basis functions of resolution levels having amplitude frequency characteristics resembling amplitude-frequency characteristics of radiation interaction data derived by irradiating calibration samples having known concentration of the analyte.

7. A method according to claim 6, further comprising irradiating calibration samples and applying wavelet basis functions to calibration data derived from the irradiation over a spectrum of frequencies of the calibration samples to derive the parameters.

8. A method according to claim 7, further comprising selecting wavelet basis functions having prominent features over a selected portion of a spectrum of frequencies of the electromagnetic radiation irradiating the sample corresponding to frequency and spatial characteristics of the radiation interaction of the analyte.

9. A method according to claim 7, further comprising selecting wavelet basis functions and resolution levels thereof having prominent peaks over a spectrum of frequencies resembling frequency and spatial characteristics of the radiation interaction of the analyte.

10. A method according to claim 7, comprising forming the modeling algorithm by deriving a regression model having a vectorial equation the constants of which are obtained by applying wavelet basis functions to calibration radiation interaction data over a spectrum of frequencies.

11. A method according to claim 10 wherein the regression model has the form of y_pred =b_o +x'"'"'_i b where y_pred is a vector of the predicted concentration, b_o is a scalar constant and b is a vector of constants derived by applying wavelet basis functions, and where x'"'"'_i is a vector of radiation interaction analytical data;

b_o and b being the parameters in the modeling algorithm.

12. A method according to claim 7, further comprising analyzing the spectra of the radiation interaction of the calibration sample to derive the calibration data and performing wavelet transform on the calibration data to obtain wavelet basis functions of the calibration data.

13. A method according to claim 7, wherein the step of irradiating calibration samples includes irradiating near infrared radiation and the method further comprises applying the wavelet basis functions by selecting Daubechies sets corresponding to frequency and amplitude characteristics of radiation interaction of glucose.

14. A method according to claim 13, wherein the Daubechies sets are selected based on wavelet basis function resolution levels having prominent features corresponding to frequency and amplitude characteristics of absorbance spectra of glucose.

15. A method according to claim 6, wherein step (a) comprises directing electromagnetic waves of frequencies that penetrate non-invasively a physiological body surface to irradiate tissue underneath the body surface to result in the radiation detected.

16. A method according to claim 6, further comprising truncating part of the data of step (b) to reduce noise before applying wavelet analysis.

17. A method according to claim 6, further comprising the step of irradiating a spectrum of electromagnetic radiation on multiple calibration samples to result in radiation interaction and detecting electromagnetic radiation resulting from the radiation interaction from the calibration samples to result in the calibration data for developing the modeling algorithm, said detected electromagnetic radiation having a signal indicative of the analyte in the calibration samples.

18. A method for non-invasively detecting quantitatively an analyte in a physiological fluid-containing sample from a patient, comprising:

(a) irradiating electromagnetic radiation at a surface of the physiological sample to result in radiation interaction of the electromagnetic radiation with physiological fluid in the sample;

(b) detecting electromagnetic radiation resulting from the radiation interaction from the sample to result in evaluation data, said detected electromagnetic radiation having a signal indicative of the analyte in the physiological fluid in the sample; and

(c) applying a modeling algorithm to at least a portion of the evaluation data to determine the quantitative characteristics of the analyte in the physiological fluid based on the evaluation data, the modeling algorithm having been developed digitally by applying wavelet basis functions of resolution levels having amplitude-frequency characteristics resembling amplitude-frequency characteristics of calibration data of radiation interaction derived by shining electromagnetic radiation of a spectrum of frequencies on calibration samples having known concentration of the analyte.

19. A method according to claim 18, further comprising:

(a) irradiating a spectrum of electromagnetic radiation on multiple calibration samples to result in radiation interaction and detecting electromagnetic radiation resulting from the radiation interaction from the calibration samples to result in the calibration data, said detected electromagnetic radiation having a signal indicative of the analyte in the calibration samples; and

(b) digitally processing at least a portion of the calibration data in matrix form covering the multiple calibration samples under the spectrum using wavelet-basis-functions to derive the model algorithm of the radiation interaction of the analyte via selecting wavelet basis functions and resolution levels thereof having prominent peaks in the amplitude-frequency characteristics resembling amplitude-frequency characteristics of the radiation interaction of the analyte.

20. An article of manufacture comprising a program storage medium, tangibly embodying a program code means readable by a computer for causing the computer to analyze the concentration of an analyte in a sample, the program code means including:

(a) code means for digitally processing calibration data obtained from irradiation of a plurality of calibration samples with electromagnetic radiation of a spectrum of frequencies, said means for digitally processing applying wavelet basis functions of resolution levels having amplitude-frequency characteristics resembling the amplitude-frequency characteristics of the calibration data to derive a model algorithm of the radiation interaction of the calibration sample; and

(b) code means for applying the modeling algorithm in matrix operation to the evaluation data obtained from radiating evaluation samples with electromagnetic radiation of a spectrum of frequencies, to determine the quantitative characteristics of the analyte in the evaluation sample based on the evaluation data.