Systems and methods for identifying classes of substances

Patent metadata

Field	Value
Publication number	US-9244045-B2
Application number	US-201414218077-A
Country	US
Kind code	B2
Filing date	Mar 18, 2014
Priority date	Apr 16, 2010
Publication date	Jan 26, 2016
Grant date	Jan 26, 2016

Abstract

In one embodiment, a system and a method for identifying the class of a component of a mixture includes collecting samples from a sample source, determining a summed ion spectrum for each sample and generating sample data from the summed ion spectra, comparing the sample data with reference summed ion spectra of multiple reference substances to determine correlations between the reference substances and the sample data, and evaluating the correlations of the substances of each substance class to determine which substance class most closely correlates to the sample data.

First claim

We claim:

1. A method for identifying a class of a component of a mixture in a sample, the method performed by a computer-based system having one or more I/O device and an instruction execution system, the method comprising:
receiving ion intensity information obtained from the component of the mixture in the sample by the one or more I/O device;
determining, by the instruction execution system, summed ion spectra from the ion intensity information;
generating, by the instruction execution system, sample data from the summed ion spectra;
comparing, by the instruction execution system, the sample data with reference summed ion spectra of multiple reference substances to determine correlations between the reference substances and the sample data, each reference substance belonging to a particular substance class; and
identifying, by the instruction execution system, the class of the component of the mixture in the sample by evaluating the correlations of substance classes to determine which substance class most closely correlates to the sample data.

2. The method of claim 1 , wherein the sample is debris from a scene of at least one of a fire and an explosion.

3. The method of claim 1 , wherein summed ion spectra of the summed ion spectrum comprise intensities at different mass-to-charge ratios.

4. The method of claim 1 , wherein comparing the sample data with the reference summed ion spectra comprises performing target factor analysis.

5. The method of claim 4 , wherein the sample includes multiple samples and wherein performing target factor analysis comprises compiling the summed ion spectra of the samples into a data matrix and performing principal components analysis on the data matrix to represent the data matrix as the product of a scores matrix and a loading matrix.

6. The method of claim 5 , wherein the reference summed ion spectra are test vectors and performing target factor analysis further comprises selecting one of the scores matrix and a loading matrix and separately transforming the selected matrix relative to each test vector to obtain a transformation vector for each test vector.

7. The method of claim 6 , wherein performing target factor analysis further comprises individually multiplying the transformation vectors and the selected matrix to obtain an associated predicted vector for each test vector and comparing the test vector with its associated predicted vector to determine how closely they correlate.

8. The method of claim 1 , wherein comparing, by the instruction execution system, the sample data with reference summed ion spectra of multiple reference substances generates comparisons, and wherein sets of correlations are obtained from the comparisons, one set for each substance class, and wherein determining which substance class most closely correlates to the sample data comprises determining which set of correlations correlates most closely to the sample data using Bayesian decision theory.

9. A system for identifying a class of a component of a mixture, the system comprising:
a processing device; and
memory accessible by the processing device, the memory storing a substance classification system that when executed by the processing device is configured to cause the processing device to:
obtain sample data associated with summed ion spectra of samples of the mixture via at least one of a local interface and an I/O device,
compare the sample data with reference summed ion spectra of multiple reference substances to determine correlations between the reference substances and the sample data, each reference substance belonging to a particular substance class, and
identify the class of the component of the mixture by evaluating the correlations of the substances of each substance class to determine which substance class most closely correlates to the sample data.

10. The system of claim 9 , wherein the substance classification system is configured to compare the sample data with the reference summed ion spectra by performing target factor analysis.

11. The system of claim 10 , wherein performing target factor analysis comprises compiling the summed ion spectra of the samples into a data matrix and performing principal components analysis on the data matrix to represent the data matrix as the product of a scores matrix and a loading matrix.

12. The system of claim 11 , wherein the reference summed ion spectra are test vectors and performing target factor analysis further comprises selecting one of the scores matrix and a loading matrix and separately transforming the selected matrix relative to each test vector to obtain a transformation vector for each test vector.

13. The system of claim 12 , wherein performing target factor analysis further comprises individually multiplying the transformation vectors and the selected matrix to obtain an associated predicted vector for each test vector and comparing the test vector with its associated predicted vector to determine how closely they correlate.

14. The system of claim 9 , wherein comparing the sample data with reference summed ion spectra of multiple reference substances generates comparisons, and wherein sets of correlations are obtained from the comparisons, one set for each substance class, and wherein the substance classification system determines which substance class most closely correlates to the sample data by determining which set of correlations correlates most closely to the sample data using Bayesian decision theory.

15. A computer-readable medium that stores a substance classification system, the substance classification system comprising:
logic configured to obtain summed ion spectra via at least one of a local interface and an I/O device and generate sample data from the summed ion spectra;
logic configured to compare sample data associated with summed ion spectra of samples of the mixture with reference summed ion spectra of multiple reference substances to determine correlations between the reference substances and the sample data, each reference substance belonging to a particular substance class; and
logic configured to evaluate the correlations of the substances of each substance class to determine which substance class most closely correlates to the sample data.

16. The computer-readable medium of claim 15 , wherein the logic configured to compare is configured to compare the sample data with the reference summed ion spectra by performing target factor analysis.

17. The computer-readable medium of claim 16 , wherein performing target factor analysis comprises compiling the summed ion spectra into a data matrix and performing principal components analysis on the data matrix to represent the data matrix as the product of a scores matrix and a loading matrix.

18. The computer-readable medium of claim 17 , wherein the reference summed ion spectra are test vectors and performing target factor analysis further comprises selecting one of the scores matrix and a loading matrix and separately transforming the selected matrix relative to each test vector to obtain a transformation vector for each test vector.

19. The computer-readable medium of claim 18 , wherein performing target factor analysis further comprises individually multiplying the transformation vectors and the selected matrix to obtain an associated predicted vector for each test vector and comparing the test vector with its associated predicted vector to determine how closely they correlate.

20. The computer-readable medium of claim 15 , wherein sets of correlations are o

Assignees

• Univ Central Florida Res Found

Inventors

• Sigman Michael
• Williams Mary

Classifications

• G01N31/00 Primary

  Investigating or analysing non-biological materials by the use of the chemical methods specified in the subgroup; Apparatus specially adapted for such methods · CPC title

• G06F2218/12

  Classification; Matching · CPC title

• G01N30/7206 Primary

  interfaced to gas chromatograph (interfaces in general for introducing or extracting samples to be analysed with specially adapted mass spectrometer, see H01J49/04) · CPC title

• G06F18/2135

  based on approximation criteria, e.g. principal component analysis · CPC title

• G06K9/00536

  Physics · mapped topic