Method of processing noninvasive spectra

• US 7,640,140 B2
• Filed: 03/30/2005
• Issued: 12/29/2009
• Est. Priority Date: 03/07/2003
• Status: Active Grant

• Current Assignee: Willow Laboratories Incorporated
• Original Assignee: Sensys Medical, Inc. (Sensys Medical Limited)
• Inventors: Ruchti, Timothy L., Blank, Thomas B., Lorenz, Alexander D.
• Primary Examiner(s): Bui; Bryan
• Application Number:
  US11/095,331
• Publication Number:
  US 20050240090A1
• Time in Patent Office: 1,735 Days
• Field of Search: 702/19, 702/189, 600/316, 600/317, 600/332
• US Class Current: 702/189
• CPC Class Codes: A61B 5/14532 for measuring glucose, e.g. by tissue impedance measurement

Abstract

This invention provides a method and apparatus that corrects for tissue related interference calibration and/or measurement of biological parameters noninvasively. The invention concerns such terms as outlier identification, filtering, spectral correction, and baseline subtraction steps that, when used together, provides for noninvasive measurement of biological parameters, such as glucose concentration.

Claims

1. A method for processing near-infrared tissue measurements to provide a calibration model for noninvasive estimation of a biological parameter, comprising the steps of:

noninvasively collecting a set of spectra with an analyzer;

removing gross outliers from said spectra to yield outlier removed spectra;

processing said outlier removed spectra to yield enhanced spectra;

removing background interference from said enhanced spectra to yield background corrected spectra; and

generating a calibration model using said background corrected spectra,

wherein said step of processing comprises the step of;

filtering said spectra with any of;

an infinite-impulse response filter; and

a finite-impulse response filter

wherein said step of filtering comprises use of any of;

a low-pass filter; and

a high-pass filter, and

wherein said high-pass filter comprises a break frequency determined via any of;

a harmonic analysis of net analysis signal; and

empirically through an exemplary set of data.

2. The method of claim 1 wherein said step of removing gross outliers comprises the steps of detecting and removing spectra having any of:

instrument errors; and

sampling errors.

3. The method of claim 2, wherein said instrument errors are detected based upon performance specifications comprising any of:

noise;

signal level;

wavelength accuracy;

wavelength precision;

stray light; and

measurement stability.

4. The method of claim 2, wherein said sampling errors are detected based upon any of:

imprecision;

tissue transients;

tissue distortion; and

poor surface contact.

5. The method of claim 1, wherein said step of filtering is optimized using a net analyte signal of said biological parameter.

6. The method of claim 1, wherein said low-pass filter maximizes a ratio of net analyte signal of said biological parameter to noise.

7. The method of claim 6, wherein said biological parameter comprises glucose concentration.

8. The method of claim 1, wherein said step of processing further comprises the step of:

correcting said spectra using at least one of;

a net analyte signal of said biological parameter;

a scaling factor; and

an offset correction.

9. The method of claim 8, wherein said step of filtering precedes said step of correcting.

10. The method of claim 9, wherein said step of processing further comprises a step of wavelength selection.

11. The method of claim 10, wherein said step of removing background interference comprises removal of any of:

a spectral background; and

a tissue template; and

a mean of a data cluster.

12. The method of claim 1, wherein said step of removing background interference comprises removal of a tissue template.

13. The method of claim 1, further comprising the step of:

performing enhanced outlier detection and removal of outlier spectra on said background corrected spectra to yield processed spectra.

14. The method of claim 13, wherein said step of performing enhanced outlier detection comprises the step of:

comparing said background corrected spectra to a specification through any of;

a distance measurement;

a cluster analysis; and

a principal component analysis.