Spectroscopic method and apparatus for optically detecting abnormal mammalian epithelial tissue

• US 6,095,982 A
• Filed: 12/11/1997
• Issued: 08/01/2000
• Est. Priority Date: 03/14/1995
• Status: Expired due to Term

• Current Assignee: Board of Regents of the University of Texas System (University of Texas System)
• Original Assignee: Board of Regents of the University of Texas System (University of Texas System)
• Inventors: Ramanujam, Nirmala, Mahadevan, Anita, Richards-Kortum, Rebecca, Mitchell, Michele Follen
• Primary Examiner(s): Lateef, Marvin M.
• Assistant Examiner(s): SHAW, SHAWNA JEANNINE
• Application Number:
  US08/988,840
• Time in Patent Office: 964 Days
• Field of Search: 600/407, 600/310, 600/342, 600/473, 600/475, 600/476, 600/477, 600/478, 250/461.2, 250/459.1, 606/3, 606/14-16, 356/301, 356/303, 356/318, 607/88-93
• US Class Current: 600/476
• CPC Class Codes: A61B 5/0071 by measuring fluorescence emission
• CPC Class Codes: A61B 5/0075 by spectroscopy, i.e. measuring spectra, e.g. Raman spectroscopy, infrared absorption spectroscopy A61B5/0071 takes precedence
• CPC Class Codes: A61B 5/0084 for introduction into the body, e.g. by catheters A61B5/1459 takes precedence
• CPC Class Codes: A61B 5/0086 using infrared radiation
• CPC Class Codes: A61B 5/4331 of the cervix A61B5/435 takes precedence
• CPC Class Codes: A61B 5/7264 Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems neural networks per se G06N3/00; expert systems per se G06N5/00
• CPC Class Codes: G01N 2021/656 Raman microprobe
• CPC Class Codes: G01N 21/6486 Measuring fluorescence of biological material, e.g. DNA, RNA, cells G01N21/6428 takes precedence
• CPC Class Codes: G01N 21/65 Raman scattering
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

A method and apparatus for detecting tissue abnormality, particularly precancerous cervical tissue, through fluorescence or Raman spectroscopy, or a combination of fluorescence and Raman spectroscopy. In vivo fluorescence measurements were followed by in vitro NIR Raman measurements on human cervical biopsies. Fluorescence spectra collected at 337, 380 and 460 nm excitation were used to develop a diagnostic method to differentiate between normal and dysplastic tissues. Using a fluorescence diagnostic method, a sensitivity and specificity of 80% and 67% were observed for differentiating squamous intraepithelial lesions (SILs) from all other tissues. In accordance with another aspect of the invention, using Raman scattering peaks observed at selected wavenumbers, SILs were separated from other tissues with a sensitivity and specificity of 88% and 100%. In addition, inflammation and metaplasia samples are correctly separated from the SILs.

Claims

1. A method of detecting and quantifying abnormal tissue in a tissue sample, comprising:

providing a tissue sample;

illuminating the tissue sample with electromagnetic radiation of a first wavelength selected to cause said tissue sample to produce a fluorescence intensity spectrum indicative of a first tissue abnormality property;

detecting a first fluorescence intensity spectrum emitted from said tissue sample as a result of illumination with said first wavelength electromagnetic radiation;

calculating from the first fluorescence intensity spectrum a probability of the tissue sample having the first tissue abnormality property;

illuminating the tissue sample with electromagnetic radiation of a second wavelength selected to cause said tissue sample to produce a fluorescence intensity spectrum indicative of a second tissue abnormality property;

detecting a second fluorescence intensity spectrum emitted from said tissue sample as a result of illumination with said second wavelength electromagnetic radiation;

calculating from the second fluorescence intensity spectrum a probability of the tissue sample having the second tissue abnormality property; and

quantifying the tissue sample from the first and second tissue abnormality property probability calculating steps.

2. A method as in claim 1, wherein:

the tissue sample is cervical tissue;

the first wavelength is about 337 nm and the first tissue abnormality property is squamous intraepithelial lesion as distinguished from normal squamous epithelial tissue; and

the second wavelength is about 380 nm and the second tissue abnormality property is squamous intraepithelial lesion as distinguished from normal columnar epithelia and inflammation.

3. A method as in claim 1, wherein:

the tissue sample is cervical tissue;

the first wavelength is about 460 nm and the first tissue abnormality property is squamous intraepithelial lesion as distinguished from normal squamous epithelial tissue; and

the second wavelength is about 380 nm and the second tissue abnormality property is squamous intraepithelial lesion as distinguished from normal columnar epithelia and inflammation.

4. A method as in claim 1, wherein:

the tissue sample is cervical tissue;

the first wavelength is further selected to cause said tissue sample to produce a fluorescence intensity spectrum indicative of a third tissue abnormality property;

the first wavelength is about 460 nm, the first tissue abnormality property is squamous intraepithelial lesion as distinguished from normal squamous epithelial tissue, and the third tissue abnormality property is high grade squamous intraepithelial lesion as distinguished from low grade squamous intraepithelial lesion; and

the second wavelength is about 380 nm and the second tissue abnormality property is squamous intraepithelial lesion as distinguished from normal columnar epithelia and inflammation.

5. A method as in claim 1, further comprising:

illuminating the tissue sample with electromagnetic radiation of a third wavelength selected to cause said tissue sample to produce a fluorescence intensity spectrum indicative of a third tissue abnormality property;

detecting a third fluorescence intensity spectrum emitted from said tissue sample as a result of illumination with said second wavelength electromagnetic radiation; and

calculating from the third fluorescence intensity spectrum a probability of the tissue sample having the third tissue abnormality property;

wherein the quantifying step further comprises qualifying the tissue sample from the third tissue abnormality property probability calculating step in addition to the first and second tissue abnormality property probability calculating steps.

6. A method as in claim 5, wherein:

the tissue sample is cervical tissue;

the first wavelength is about 337 nm and the first tissue abnormality property is squamous intraepithelial lesion as distinguished from normal squamous epithelial tissue;

the second wavelength is about 380 nm and the second tissue abnormality property is squamous intraepithelial lesion as distinguished from normal columnar epithelia and inflammation; and

the third wavelength is about 460 nm and the third tissue abnormality property is high grade squamous intraepithelial lesion as distinguished from low grade squamous intraepithelial lesion. sample belongs to the particular tissue category.

7. A method of probabilistically classifying a sample of tissue, comprising:

illuminating the tissue sample with electromagnetic radiation having wavelengths of about 337 nm, about 380 nm, and about 460 nm to excite the tissue sample into producing first, second and third fluorescence intensity spectra, respectively;

detecting the first, second and third fluorescence intensity spectra to obtain respective first, second and third spectral data; and

calculating a probability that the tissue sample belongs to a first tissue classification from at least one of the first and third spectral data, a probability that the tissue sample belongs to the second tissue classification from the second spectral data, and a probability that the tissue sample belongs to the third tissue classification from the third spectral data.

8. A method as in claim 7, further comprising:

normalizing and mean-scaling the first spectral data prior to the calculating step;

normalizing and mean-scaling the second spectral data prior to the calculating step; and

normalizing the third spectral data prior to the calculating step.

9. A method of probabilistically classifying a sample of tissue, comprising:

illuminating the tissue sample with electromagnetic radiation having wavelengths of about 380 nm and about 460 nm to excite the tissue sample into producing first, second and third fluorescence intensity spectra, respectively;

detecting the first and second fluorescence intensity spectra to obtain respective first and second spectral data; and

calculating a probability that the tissue sample belongs to a first tissue classification from the second spectral data, a probability that the tissue sample belongs to the second tissue classification from the first spectral data, and a probability that the tissue sample belongs to the third tissue classification from the second spectral data.

10. A method as in claim 9, further comprising:

normalizing the second spectral data prior to the calculating step;

normalizing and mean-scaling the first spectral data prior to the calculating step; and

mean-scaling the normalized second spectral data prior to the calculating step.

11. A method of developing an index for calculating a probability that a tissue belongs to one of a plurality of histo-pathologic tissue classifications, comprising:

providing a plurality of tissue samples;

illuminating the tissue samples with electromagnetic radiation having wavelengths of about 337 nm, about 380 nm, and about 460 nm to excite the tissue samples into producing fluorescence intensity spectra;

detecting the fluorescence intensity spectra to obtain first, second and third spectral data from the tissue samples illuminated at respectively the wavelengths of about 337 nm, about 380 nm and about 460 nm in the illuminating step; and

forming a first set of principle components from the first spectral data that account for a significant amount of variation in the first spectral data and show statistically significant differences between first and second histo-pathologic tissue classifications;

calculating first probability distribution functions for the first and second histo-pathologic tissue classifications;

forming a second set of principle components from the second spectral data that account for a significant amount of variation in the second spectral data and show statistically significant differences between third and fourth histo-pathologic tissue classifications;

calculating second probability distribution functions for the third and fourth histo-pathologic tissue classifications;

forming a third set of principle components from the third spectral data that account for a significant amount of variation in the third spectral data and show statistically significant differences between fifth and sixth histo-pathologic tissue classifications;

calculating third probability distribution functions for the fifth and sixth histo-pathologic tissue classifications.

12. A method of assigning a probability that a tissue sample belongs to a particular tissue category, comprising:

providing a tissue sample;

illuminating the tissue sample with electromagnetic radiation having wavelengths of about 337 nm, about 380 nm, and about 460 nm to excite the tissue sample into producing fluorescence intensity spectra;

detecting the fluorescence intensity spectra to obtain spectral data;

obtaining principal components PC1, PC2, PC4, PC5, and PC7 from said spectral data, and

calculating from principal components PC1 and PC2 from spectral data at about 337 nm a probability of squamous intraepithelial lesion as distinguished from normal squamous epithelial tissue, from principal components PC2 and PC5 from spectral data at about 380 rn a probability of squamous intraepithelial lesion as distinguished from normal columnar epithelia and inflammation, and from principal components PC4 and PC7 from spectral data at about 460 nm a probability of high grade squamous intraepithelial lesion as distinguished from low grade squamous intraepithelial lesion.

13. A method of assigning a probability that a tissue sample belongs to a particular tissue category, comprising:

providing a tissue sample;

illuminating the tissue sample with electromagnetic radiation having wavelengths of about 380 nm and about 460 nm to excite the tissue sample into producing fluorescence intensity spectra;

detecting the fluorescence intensity spectra to obtain spectral data;

obtaining principal components PC1, PC2, PC4, PC5, and PC7 from said spectral data; and

calculating from principal components PC1 and PC2 from spectral data at about 460 nm a probability of squamous intraepithelial lesion as distinguished from normal squamous epithelial tissue, from principal components PC2 and PC5 from spectral data at about 380 nm a probability of squamous intraepithelial lesion as distinguished from normal columnar epithelia and inflammation, and from principal components PC4 and PC7 from spectral data at about 460 nm a probability of high grade squamous intraepithelial lesion as distinguished from low grade squamous intraepithelial lesion.

14. A method of detecting abnormal epithelial tissue in or from an anatomical feature of a mammal, comprising:

identifying a plurality of individuals having the anatomical feature from a population of interest, the respective anatomical features of the individuals collectively having a diversity of epithelial tissue categories, including normal and abnormal epithelial tissue categories, that is representative of the diversity of the epithelial tissue categories collectively occurring in the respective anatomical features of the population of interest;

optically applying electromagnetic energy comprising a first and second wavelengh to plural epithelial tissue regions of the respective anatomical features of the individuals, said first and second wavelengths selected to produce spectra indicative of a first and second tissue abnormality property, respectively;

optically detecting spectra from the epithelial tissue regions of the respective anatomical features of the individuals resulting from the step of optically applying electromagnetic energy to plural epithelial tissue regions;

processing the optically detected spectra into first spectral data;

transforming the first spectral data into a dimensionally reduced set of variables that significantly accounts for variance in the first spectral data and that exhibits statistically significant differences between the epithelial tissue categories;

optically applying electromagnetic energy comprising said first and second wavelength to epithelial tissue of the anatomical feature of a subject individual, said first and second wavelengths selected to produce spectra indicative of said first and second tissue abnormality properties, respectively;

optically detecting a spectrum from the epithelial tissue of the anatomical feature of the subject individual resulting from the step of optically applying electromagnetic energy to epithelial tissue;

processing the optically detected spectrum into second spectral data; and

calculating the probability that the epithelial tissue of the anatomical feature of the subject individual belongs to each of one or more of the epithelial tissue categories from the second spectral data and the dimensionally reduced set.

15. A method as in claim 14, wherein:

each of the epithelial tissue categories has a prior probability of occurring in the individuals;

the first spectral data transforming step comprises calculating principal components and principle component scores; and

the probability calculating step comprises classifying the epithelial tissue of the anatomical feature of the subject individual in the tissue category for which a posterior probability is highest, the posterior probability being calculated from the prior probability, the second spectral data, the principal components, and the principal component scores.

16. A method as in claim 15, further comprising:

preprocessing the first spectral data prior to the first spectral data transforming step to remove intra-patient and inter-patient variation; and

preprocessing the second spectral data prior to the probability calculating step to remove intra-patient variation.

17. A method as in claim 14, wherein the optically detected spectra and the optically detected spectrum comprise fluorescence spectra.

18. A method as in claim 17, wherein:

the anatomical feature is a human cervix;

the epithelial tissue categories are squamous intraepithelial lesion and normal squamous epithelial tissue; and

the electromagnetic energy comprises a wavelength of about 337 nm.

19. A method as in claim 18, further comprising the step of normalizing and mean-scaling the first spectral data prior to the transforming step.

20. A method as in claim 17, wherein:

the anatomical feature is a human cervix;

the epithelial tissue categories are squamous intraepithelial lesion and normal columnar epithelia and inflammation; and

the electromagnetic energy comprises a wavelength of about 380 nm.

21. A method as in claim 20, further comprising the step of normalizing and mean-scaling the first spectral data prior to the transforming step.

22. A method as in claim 17, wherein:

the anatomical feature is a human cervix;

the epithelial tissue categories are squamous intraepithelial lesion and normal squamous epithelial tissue; and

the electromagnetic energy comprises a wavelength of about 460 nm.

23. A method as in claim 22, further comprising the step of normalizing and mean-scaling the first spectral data prior to the transforming step.

24. A method as in claim 17, wherein:

the anatomical feature is a human cervix;

the epithelial tissue categories are high grade squamous intraepithelial lesion and low grade squamous intraepithelial lesion; and

the electromagnetic energy comprises a wavelength of about 460 nm.

25. A method as in claim 24, further comprising the step of normalizing the first spectral data prior to the transforming step.

26. A method as in claim 17, wherein:

the anatomical feature is a human cervix;

the epithelial tissue categories are high grade squamous intraepithelial lesion, low grade squamous intraepithelial lesion, squamous intraepithelial lesion, normal squamous epithelial tissue, normal columnar epithelia, and inflammation; and

the electromagnetic energy comprises wavelengths of about 337 nm, 380nm, and 460 nm.

27. A method of identifying a probable tissue category for epithelial tissue of a mammalian patient from among a plurality of epithelial tissue categories, comprising:

identifying the patient with a predetermined population having prior probabilities of the tissue categories therein;

applying electromagnetic radiation comprising a first and second wavelength to a plurality of tissue sites of subjects in the population and to the tissue of the patient, the first and second wavelengths selected to excite respectively a first autofluorescence intensity spectra indicative of first and second tissue categories and second autofluorescence intensity spectra indicative of third and fourth tissue categories;

obtaining first and second sets of subject autofluorescence spectral data from the first and second autofluorescence intensity spectra from the electromagnetic radiation applying step;

preprocessing the first and second sets of subject autofluorescence spectral data to reduce inter-subject and intra-subject variation therein;

forming a first dimensionally reduced set of emission variables from the preprocessed set of first subject autofluorescence spectral data, including a first reduced eigenvector matrix, that shows statistically significant differences between the first and second tissue categories and that significantly accounts for variation in the preprocessed set of first subject autofluorescence spectral data;

calculating first subject scores from the first dimensionally reduced set of emission variables for the first and second tissue categories;

forming a second dimensionally reduced set of emission variables from the preprocessed set of second subject autofluorescence spectral data, including a second reduced eigenvector matrix, that shows statistically significant differences between the third and fourth tissue categories and that significantly accounts for variation in the preprocessed set of second subject autofluorescence spectral data,

calculating second subject scores from the second dimensionally reduced set of emission variables for the third and fourth tissue categories;

obtaining respective sets of patient autofluorescence spectral data from the electromagnetic radiation applying step;

preprocessing the sets of patient autofluorescence spectral data to reduce intra-patient variation therein;

concatenating the preprocessed patient autofluorescence spectral data into vectors;

processing the vectors with the first reduced eigenvector matrix to obtain first patient scores;

calculating posterior probabilities for the first and second tissue categories from the first subject scores, from the first patient scores, and from prior probabilities of the first and second tissue categories;

processing the vectors with the second reduced eigenvector matrix to obtain second patient scores; and

calculating posterior probabilities for the third and fourth tissue categories from the second subject scores, from the second patient scores, and from prior probabilities of the third and fourth tissue categories.

28. A method as in claim 27, wherein the step of forming a first dimensionally reduced set of emission variables comprises:

forming first principal components and first principle component scores from the preprocessed set of first subject autofluorescence spectral data;

retaining first eigenvalues from the principal components forming step that account for a significant amount of the variation in the preprocessed set of first subject autofluorescence spectral data;

calculating the diagnostic contribution of each of the first principle components for the first retained eigenvalues; and

retaining the eigenvalues corresponding to the first principle components identified in the diagnostic contribution calculating step as having a significant diagnostic contribution;

and wherein the step of forming a second dimensionally reduced set of emission variables comprises;

forming second principal components and second principle component scores from the preprocessed set of second subject autofluorescence spectral data;

retaining second eigenvalues from the principal components forming step that account for a significant amount of the variation in the preprocessed set of second subject autofluorescence spectral data;

calculating the diagnostic contribution of each of the second principle components for the second retained eigenvalues; and

retaining the eigenvalues corresponding to the second principle components identified in the diagnostic contribution calculating step as having a significant diagnostic contribution.

29. A method as in claim 28, wherein the posterior probability calculating step comprises calculating posterior probability using logistic discrimination.

30. A method as in claim 29, wherein the diagnostic contribution calculating step comprises calculating the diagnostic contribution using a Student'"'"'s T-Test.

31. A method of obtaining a probability that an epithelial tissue site of a mammalian patient contains tissue belonging to a particular tissue category, comprising:

identifying a training set of epithelial tissue sites identified with a plurality of individuals, the training set of epithelial tissue sites having tissues belonging to a plurality of tissue categories and occurring with known prior probabilities;

obtaining optical response training data from the epithelial tissue sites in the training set by illuminating said tissue sites with electromagnetic radiation comprising a first wavelength selected to produce first spectra indicative of first and second tissue categories, and a second wavelength selected to produce second spectra indicative of third and fourth tissue categories;

calculating a first dimensionally reduced set of variables from the optical response training data that shows statistically significant differences between the first and second tissue categories and that significantly accounts for variation in the first spectra;

calculating a second dimensionally reduced set of variables from the optical response training data that shows statistically significant differences between the third and fourth tissue categories and that significantly accounts for variation in the preprocessed optical response training data;

obtaining optical response patient data from the epithelial tissue site of the patient; and

calculating the probability that the epithelial tissue site of the patient contains the particular tissue type using the optical response patient data, the prior probabilities, and the first and second dimensionally reduced set.

32. A method as in claim 31, wherein:

the first dimensionally reduced set calculating step comprises transforming the first spectra into first principal components that significantly accounts for variation in the first spectra, and testing the first principal components for statistically significant differences between the first and second tissue categories using a Student'"'"'s T-Test;

the second dimensionally reduced set calculating step comprises transforming the second spectra into second principal components that significantly accounts for variation in the second spectra, and testing the second principal components for statistically significant differences between the third and fourth tissue categories using a Student'"'"'s T-Test; and

the probability calculating step comprises logistic discrimination.

33. A method as in claim 32, further comprising the step of preprocessing the optical response training data prior to the first dimensionally reduced set calculating step and the second dimensionally reduced set calculating step to remove inter-individual variations therein.

34. An index embodied on a computer-readable medium for calculating a probability that epithelial tissue in or from an anatomical feature of a mammal belongs to each of a plurality of epithelial tissue categories, the index being formed by the steps of:

identifying a plurality of individuals having the anatomical feature from a population of interest, the respective anatomical features of the individuals collectively having a diversity of the epithelial tissue categories, including normal and abnormal epithelial tissue categories, that is representative of the diversity of the epithelial tissue categories collectively occurring in the respective anatomical features of the population of interest;

optically applying electromagnetic energy comprising a first wavelength and a second wavelength to plural epithelial tissue regions of the respective anatomical features of the individuals, said first and second wavelengths selected to produce first fluorescence intensity spectra indicative of a first tissue abnormality and second fluorescence intensity spectra indicative of a second tissue abnormality, respectively;

optically detecting the first and second fluorescence intensity spectra from the epithelial tissue regions of the respective anatomical features of the individuals resulting from the step of optically applying electromagnetic energy;

processing the first and second fluorescence intensity spectra into first and second spectral data;

transforming the first spectral data into a first dimensionally reduced set of variables that significantly accounts for variance in the first spectral data and that exhibits statistically significant differences between the first tissue abnormality and at least one of the other epithelial tissue categories;

transforming the second spectral data into a second dimensionally reduced set of variables that significantly accounts for variance in the second spectral data and that exhibits statistically significant differences between the second tissue abnormality and at least one of the other epithelial tissue categories; and

deriving the index from the first and second dimensionally reduced sets.

35. An index as in claim 34 wherein the first and second spectral data transforming step comprises calculating principal components and principle component scores.

36. An index as in claim 35, further comprising preprocessing the first and second spectral data prior to the first and second spectral data transforming steps to remove intra-patient and inter-patient variation.

37. An index as in claim 34, wherein the optically detected spectra comprises fluorescence spectra.

38. An index as in claim 37, wherein:

the anatomical feature is a human cervix;

the epithelial tissue categories are squamous intraepithelial lesion and normal squamous epithelial tissue; and

the electromagnetic energy comprises a wavelength of about 337 nm.

39. An index as in claim 38, further comprising the step of normalizing and mean-scaling the first spectral data prior to the transforming step.

40. An index as in claim 37, wherein:

the anatomical feature is a human cervix;

the epithelial tissue categories are squamous intraepithelial lesion and normal columnar epithelia and inflammation; and

the electromagnetic energy comprises a wavelength of about 380 nm.

41. An index as in claim 40, further comprising the step of normalizing and mean-scaling the first spectral data prior to the transforming step.

42. An index as in claim 37, wherein:

the anatomical feature is a human cervix;

the epithelial tissue categories are squamous intraepithelial lesion and normal squamous epithelial tissue; and

the electromagnetic energy comprises a wavelength of about 460 nm.

43. An index as in claim 42, further comprising the step of normalizing and mean-scaling the first spectral data prior to the transforming step.

44. An index as in claim 37, wherein:

the anatomical feature is a human cervix;

the epithelial tissue categories are high grade squamous intraepithelial lesion and low grade squamous intraepithelial lesion; and

the electromagnetic energy comprises a wavelength of about 460 nM.

45. An index as in claim 44, further comprising the step of normalizing the first spectral data prior to the transforming step.

46. An index as in claim 37, wherein:

the anatomical feature is a human cervix;

the epithelial tissue categories are high grade squamous intraepithelial lesion, low grade squamous intraepithelial lesion, squamous intraepithelial lesion, normal squamous epithelial tissue, normal columnar epithelia, and inflammation; and

the electromagnetic energy comprises wavelengths of about 337 nm, 380nm, and 460 nm.