Noise tolerant measurement

• US 20100063370A1
• Filed: 09/09/2009
• Published: 03/11/2010
• Est. Priority Date: 09/10/2008
• Status: Active Grant

• Current Assignee: Compact Imaging, Inc.
• Original Assignee: Gaston D. Baudat, Josh N. Hogan
• Inventors: Baudat, Gaston D., Hogan, Josh N.
• Granted Patent: US 8,577,432 B2
• Time in Patent Office: Days
• US Class Current: 600/316
• CPC Class Codes: A61B 5/0066 Optical coherence imaging
• 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 5/726 using Wavelet transforms
• CPC Class Codes: G16H 50/50 for simulation or modelling of medical disorders

Abstract

The invention relates to statistical methods for fitting a mathematical model of the interaction of signals, such as optical signals, with tissue to detected signals or data related to the interaction. In particular it relates to statistical methods for fitting a mathematical model of the interaction of optical signals with tissue to actual signals, such as interferometric signals related to data optical scattering in tissue and their relationship to glucose concentration. It also relates to statistical methods for fitting a mathematical model of the interaction of optical signals with tissue to data such as the spectral distribution values of optical signals absorbed or scattered by tissue and their relationship to glucose concentration. The invention provides a practical non-invasive glucose measurement method and system. The invention also provides a measurement method and system that performs well in low signal-to-noise environments.

Claims

1. A method for processing signals detected by a non-invasive analysis system to determine the value of at least one attribute of a target in a low signal to noise environment, said method comprising:

generating a system model that represents the interaction of radiation scattered by said target and at least one reference signal associated with said non-invasive analysis system;

generating a noise model, said model comprised of noise associated with the non-invasive analysis system (system noise), noise associated with the target (target noise) and noise associated with the said scattered radiation (radiation noise);

generating a formula with a number of parameters, at least one of which is related to the value of said attribute of said target, said value to be determined;

using estimation techniques to fit said formula to signals detected by said non-invasive analysis system to determine the value of said attribute of said target; and

outputting the value of said attribute of said target.

2. The method of claim 1, wherein the values of two or more attributes are determined.

3. The method of claim 1, wherein the estimation techniques includes any of the following:

maximum of likelihood techniques;

least mean square;

weighted least mean square.

4. The method of claim 1, wherein the estimation techniques used to fit said formula involve noise statistics.

5. The method of claim 1, wherein the target may include any of the following:

tissue;

tissue fluid;

interstitial fluid;

blood.

6. The method of claim 1, wherein the attribute of said target is glucose concentration.

7. The method of claim 1, wherein the non-invasive analysis system is a multiple reference analysis system.

8. The method of claim 1, wherein the non-invasive analysis system is a conventional OCT system.

9. The method of claim 1, wherein the signals detected by the non-invasive analysis system are interference signals.

10. The method of claim 1, wherein said detected signals are subjected to processing prior to using estimation techniques to fit said formula to said detected signals.

11. The method of claim 1, wherein said estimation technique is a maximum likelihood estimation which uses a constrained optimization process to fit signals detected by the non-invasive analysis system to determine the value of said attribute of said target.

12. A non-invasive analysis system comprising:

an actual analysis system 101, said actual analysis system outputting at least one actual signal 103, where said actual signals containing information obtained from a target of interest;

a system model, where said system model outputs at least one theoretical signal;

a noise model 109, where said noise model output noise statistics;

a parametric estimation processor 113, where said parametric estimation processor receives actual signals, theoretical signals and noise statistics and where said parametric estimation processor uses at least one fitness criterion to estimate at least one model parameter 115 of said system model to optimize the fit of said theoretical signal with said actual signal;

and

an output, where said output provides information about said target of interest.

13. The system as in claim 12, wherein the system model is dynamically selected.

14. The system as in claim 12, wherein the system model is generated from iterative outputs of the parametric estimation processor.

15. The system as in claim 12, wherein the system model is generated from iterative outputs of the parametric estimation processor wherein the iterative outputs include outputs which incorporate signals from multiple targets.

16. The system as in claim 12, wherein the system model is empirically generated by analyzing data sets, such that a pattern is found dynamically without being predicated on the operative physics.