Forensic integrated search technology with instrument weight factor determination

• US 8,112,248 B2
• Filed: 01/22/2008
• Issued: 02/07/2012
• Est. Priority Date: 06/09/2005
• Status: Active Grant

• Current Assignee: Chemimage Corporation
• Original Assignee: Chemimage Corporation
• Inventors: Schweitzer, Robert, Treado, Patrick J., Neiss, Jason
• Primary Examiner(s): Barbee, Manuel L
• Application Number:
  US12/017,445
• Publication Number:
  US 20080300826A1
• Time in Patent Office: 1,477 Days
• Field of Search: 702/22, 702/27, 702/28, 702 30- 32, 702/85, 702/104, 702/181, 436/43, 253/390.9, 252/521, 707/600, 707/601
• US Class Current: 702/181
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 2218/12 Classification; Matching
• CPC Class Codes: G16C 20/20 Identification of molecular entities, parts thereof or of chemical compositions
• CPC Class Codes: G16C 20/90 Programming languages; Computing architectures; Database systems; Data warehousing
• CPC Class Codes: H01J 49/0036 Step by step routines describing the handling of the data generated during a measurement

Abstract

A system and method to search spectral databases and to identify unknown materials from multiple spectroscopic data in the databases. The methodology may be substantially automated and is configurable to determine weights to be accorded to spectroscopic data from different spectroscopic data generating instruments for improved identification of unknown materials. Library spectra from known materials are divided into training and validation sets. Initial, instrument-specific weighting factors are determined using a weight grid or weight scale. The training and validation spectra are weighted with the weighting factors and indicator probabilities for various sets of “coarse” weighting factors are determined through an iterative process. The finally-selected set of coarse weighting factors is further “fine tuned” using a weight grid with finer values of weights. The instrument-specific finer weight values may be applied to test data sets (or spectra) of an unknown material as well as to the library spectra from corresponding spectroscopic instruments. Instrument-specific weights for each class of samples may also be computed for additional customization and accuracy.

Claims

1. A method comprising configuring a processor to perform the following steps:

(a) identifying a plurality of spectroscopic instruments;

(b) for each spectroscopic instrument, obtaining an instrument-specific first plurality of spectra and an instrument-specific second plurality of spectra from a plurality of samples using said spectroscopic instrument;

(c) selecting a set of instrument-specific weight values from a first plurality of weight values, wherein said set includes an instrument-specific weight value for each of said plurality of spectroscopic instruments;

(d) performing the following for the selected set of instrument-specific weight values;

for each spectroscopic instrument, assigning a respective instrument-specific weight value from the selected set of weight values to each instrument-specific spectrum in said first and said second plurality of spectra, thereby generating an instrument-specific first plurality of weighted spectra and an instrument-specific second plurality of weighted spectra;

(e) for the selected set of instrument-specific weight values, determining a corresponding indicator probability value from said first and said second plurality of weighted spectra;

(f) repeating steps (c), (d), and (e) until a first predetermined number of different sets of instrument-specific weight values are selected, thereby obtaining a first plurality of indicator probability values;

(g) selecting a first target set of instrument-specific weight values as that set of instrument-specific weight values which corresponds to a highest indicator probability value in said first plurality of indicator probability values; and

(h) for each spectroscopic instrument, applying an instrument-specific weight value from said first target set to one or more spectra obtained using said spectroscopic instrument.

2. The method of claim 1, wherein the number of indicator probability values in said first plurality of indicator probability values is equal to said first predetermined number.

3. The method of claim 1, wherein said plurality of samples includes samples of only known materials.

4. The method of claim 1, wherein each spectrum in each said first plurality of spectra is obtained from a corresponding first location on a respective one of said plurality of samples, and wherein each spectrum in each said second plurality of spectra is obtained from a corresponding second location on a respective one of said plurality of samples, and wherein said first location is different from said second location for at least one of said plurality of samples.

5. The method of claim 1, wherein each spectrum in an instrument-specific first plurality of spectra has a unique corresponding spectrum in respective instrument-specific second plurality of spectra.

6. The method of claim 1, wherein said plurality of spectroscopic instruments includes two or more of the following:

a Raman spectrometer, a mid-infrared spectrometer, an x-ray diffractometer, a fluorescence spectrometer, an energy dispersive x-ray analyzer, and a mass spectrometer.

7. The method of claim 1, wherein step (e) includes:

determining a plurality of final probability values by comparing each spectrum in each said second plurality of weighted spectra against each spectrum in each said first plurality of weighted spectra, wherein the number of final probability values in said plurality of final probability values is equal to the number of samples in said plurality of samples; and

determining said corresponding indicator probability value for the selected set of instrument-specific weight values using the following equation;

8. The method of claim 7, wherein the step of determining said plurality of final probability values includes:

obtaining a set of scores based on said comparison of each spectrum in each said second plurality of weighted spectra against each spectrum in each said first plurality of weighted spectra, wherein each score in said set of scores indicates a likelihood of a match between one of said second plurality of weighted spectra and a corresponding one of said first plurality of weighted spectra;

calculating a set of relative probability values based on the set of scores; and

fusing all relative probability values to thereby produce said plurality of final probability values.

9. The method of claim 1, wherein step (c) includes:

selecting a corresponding weight value from said first plurality of weight values for each spectroscopic instrument in such a manner that the total of all weight values in said set of instrument-specific weight values equals to “

1”

.

10. The method of claim 1, wherein step (c) includes one of the following:

initially selecting a corresponding predetermined weight value from said first plurality of weight values for each spectroscopic instrument; and

initially randomly selecting a corresponding weight value from said first plurality of weight values for each spectroscopic instrument.

11. The method of claim 1, wherein the total number of spectra in each of said first plurality of spectra and said second plurality of spectra is equal to the total number of spectroscopic instruments in said plurality of spectroscopic instruments multiplied by the total number of samples in said plurality of samples.

12. The method of claim 1, further comprising the steps of:

(i) generating a second plurality of weight values using weight values in said first target set of instrument-specific weight values;

(j) repeating steps (c), (d), and (e) using weight values from said second plurality of weight values instead of weight values from said first, plurality of weight values and until a second predetermined number of different sets of instrument-specific weight values are selected, thereby obtaining a second plurality of indicator probability values;

(k) modifying step (g) by selecting a second target set of instrument-specific weight values instead of said first target set, wherein said second target set is that set of instrument-specific weight values which corresponds to a highest indicator probability value in said second plurality of indicator probability values; and

(l) for each spectroscopic instrument in step (h), applying an instrument-specific weight value from said second target set instead of said first target set to one or more spectra obtained using said spectroscopic instrument.

13. The method of claim 12, wherein, in step (j), a corresponding weight, value from said second plurality of weight values is selected for each spectroscopic instrument in such a manner that the total of all weight values in said set of instrument-specific weight values equals to “

1”

.

14. The method of claim 12, further comprising enabling a user to perform at least one of the following:

step (c); and

step (i).

15. A method comprising configuring a processor to perform the following steps:

(a) identifying a plurality of spectroscopic instruments;

(b) classifying a plurality of samples into a plurality of classes, wherein each class includes one or more of said plurality of samples;

(c) for each spectroscopic instrument, obtaining an instrument-specific first plurality of spectra and an instrument-specific second plurality of spectra from samples in one of said plurality of classes using said spectroscopic instrument;

(d) selecting a set of instrument-specific weight values from a plurality of weight values, wherein said set includes an instrument-specific weight value for each of said plurality of spectroscopic instruments;

(e) performing the following for the selected set of instrument-specific weight values;

for each spectroscopic instrument, assigning a respective instrument-specific weight value from the selected set of weight values to each instrument-specific spectrum in said first and said second plurality of spectra, thereby generating an instrument-specific first plurality of weighted spectra and an instrument-specific second plurality of weighted spectra for the samples in said one of said plurality of classes;

(f) for the selected set of instrument-specific weight values, determining a corresponding indicator probability value from said first and said second plurality of weighted spectra;

(g) repeating steps (d), (e), and (f) until a predetermined number of different sets of instrument-specific weight values are selected, thereby obtaining a plurality of indicator probability values;

(h) selecting a target set of instrument-specific weight values as that set of instrument-specific weight values which corresponds to a highest indicator probability value in said plurality of indicator probability values; and

(i) for each spectroscopic instrument, applying an instrument-specific weight value from said target set to one or more spectra obtained using said spectroscopic instrument from samples belonging to said one of said plurality of classes.

16. The method of claim 15, wherein step (d) includes:

selecting a corresponding weight value from said plurality of weight values for each spectroscopic instrument in such a manner that the total of all weight values in said set of instrument-specific weight values equals to 1.

17. A system comprising:

a computer executable program code, which, when executed by a processor, causes said processor to perform the following operations;

(a) recognize a plurality of spectroscopic instruments;

(b) for each spectroscopic instrument, obtain an instrument-specific first plurality of spectra and an instrument-specific second plurality of spectra from a plurality of samples using said spectroscopic instrument;

(c) select a set of instrument-specific weight values from a plurality of weight values, wherein said set includes an instrument-specific weight value for each of said plurality of spectroscopic instruments;

(d) perform the following for the selected set of instrument-specific weight values;

for each spectroscopic instrument, assign a respective instrument-specific weight value from the selected set of weight values to each instrument-specific spectrum in said first and said second plurality of spectra, thereby generating an instrument-specific first plurality of weighted spectra and an instrument-specific second plurality of weighted spectra;

(e) for the selected set of instrument-specific weight values, determine a corresponding indicator probability value from said first and said second plurality of weighted spectra;

(f) repeat operations (c), (d), and (e) until a predetermined number of different sets of instrument-specific weight values are selected, thereby obtaining a plurality of indicator probability values;

(g) select a target set of instrument-specific weight values as that set of instrument-specific weight values which corresponds to a highest indicator probability value in said plurality of indicator probability values; and

(h) for each spectroscopic instrument, apply an instrument-specific weight value from said target set to one or more spectra obtained using said spectroscopic instrument.

18. The system of claim 17, further comprising one or more of the following:

said processor configured to execute said program code; and

at least one of said plurality of spectroscopic instruments configured to be operationally coupled to said processor.

19. The system of claim 17, wherein said program code, upon execution by said processor, causes said processor to further perform the following:

select a corresponding weight value from said plurality of weight values for each spectroscopic instrument in such a manner that the total of all weight values in said set of instrument-specific weight values equals to “

1”

.

20. The system of claim 17, wherein said program code, upon execution by said processor, causes said processor to further perform the following:

enable a user to identify to said processor one or more samples in said plurality of samples to be used to obtain said instrument-specific first and second plurality of spectra.