Image classifier training
View Patent ↗Methods are disclosed that include: (a) applying a first stain to a first sample having a plurality of regions, where the first stain selectively binds to only a first subset of the regions of the first sample; (b) applying a second stain to the first sample, where the second stain binds to a second set of regions of the first sample; (c) obtaining an image of the first sample, and analyzing the image to obtain a first component image corresponding substantially only to spectral contributions from the first stain, and a second component image corresponding substantially only to spectral contributions from the second stain; and (d) training a classifier to identify regions of a second sample based on information derived from the first and second component images, the identified regions corresponding to the first subset of regions of the first sample.
1. A method, comprising:
applying a first stain to a first sample having a plurality of regions, wherein the first stain selectively binds to only a first subset of the regions of the first sample;
applying a second stain to the first sample, wherein the second stain binds to a second set of regions of the first sample;
obtaining an image of the first sample, and analyzing the image to obtain a first component image corresponding substantially only to spectral contributions from the first stain, and a second component image corresponding substantially only to spectral contributions from the second stain; and
training a classifier to identify regions of a second sample based on information derived from the first and second component images, the identified regions corresponding to the first subset of regions of the first sample.
2. The method of claim 1 , wherein the first subset of regions and the second set of regions have regions in common.
3. The method of claim 1 , wherein the second set of regions comprises at least some regions that are each adjacent to a region in the first subset.
4. The method of claim 1 , wherein the second set of regions is a subset of the regions of the first sample.
5. The method of claim 1 , wherein the second set of regions corresponds to substantially all of the first sample.
6. The method of claim 1 , wherein the first subset of regions corresponds to membranes in the first sample.
7. The method of claim 1 , wherein the second set of regions corresponds to nuclei in the first sample.
8. The method of claim 1 , wherein the second set of regions corresponds to non-specific binding sites in the first sample.
9. The method of claim 1 , wherein the information derived from the first component image comprises position information corresponding to the first subset of regions.
10. The method of claim 1 , wherein the information derived from the first component image comprises information about an amount of the first stain in the first subset of regions.
11. The method of claim 1 , wherein the information derived from the second component image comprises information about an amount of the second stain in the second set of regions.
12. The method of claim 1 , further comprising applying a third stain to the first sample, wherein the third stain selectively binds to a third subset of the regions of the first sample.
13. The method of claim 12 , wherein the first subset of regions and the third subset of regions have regions in common.
14. The method of claim 12 , wherein the third subset of regions and the second set of regions have regions in common.
15. The method of claim 12 , wherein the second set of regions comprises at least some regions that are each adjacent to a region in the third subset of regions.
16. The method of claim 12 , further comprising:
analyzing the image to obtain a third component image, wherein the third component image corresponds substantially only to spectral contributions from the third stain; and
training the classifier based on information derived from the third component image.
17. The method of claim 16 , wherein the classifier is a first classifier, and further comprising training a second classifier to identify regions of another sample, the identified regions corresponding to the third subset of regions of the first sample.
18. The method of claim 12 , wherein one of the first and second stains comprises a fluorogenic stain and the other of the first and second stains comprises a chromogenic stain.
19. The method of claim 1 , wherein the second stain comprises a counterstain.
20. The method of claim 1 , further comprising applying a fourth stain to the first sample, wherein the fourth stain binds to a fourth set of regions of the first sample.
21. The method of claim 20 , wherein the fourth set of regions is a subset of regions of the first sample.
22. The method of claim 20 , wherein the fourth set of regions corresponds to substantially all of the first sample.
23. The method of claim 20 , further comprising:
analyzing the image to obtain a fourth component image, wherein the fourth component image corresponds substantially only to spectral contributions from the fourth stain; and
training the classifier based on information derived from the fourth component image.
24. The method of claim 1 , wherein the first subset of regions of the first sample comprises one or more types of tissue.
25. The method of claim 1 , wherein the first subset of regions of the first sample comprises one or more types of cells.
26. The method of claim 1 , wherein the first stain comprises at least one member of the group consisting of immunohistochemical agents, cytokeratin, and cadherin.
27. The method of claim 1 , wherein the second stain comprises at least one member of the group consisting of immunohistochemical agents, DAB, fast red, fluorescein, rhodamine, Texas red, Cy3, Cy5, Cy5.5, Alexa dyes, quantum dots, DAPI, Hoechst blue, and hematoxylin.
28. The method of claim 1 , wherein analyzing the image comprises unmixing the image to obtain the first and second component images.
29. The method of claim 1 , wherein analyzing the image comprises separating image color channels to obtain the first and second component images.
30. The method of claim 1 , wherein a difference between a wavelength of a maximum in a fluorescence emission spectrum of the first stain and a wavelength of a maximum in a fluorescence emission spectrum of the second stain is 50 nm or more.
31. The method of claim 1 , wherein a difference between a wavelength of maximum absorption of the first stain and a wavelength of maximum absorption of the second stain is 50 nm or more.
32. The method of claim 1 , further comprising obtaining birefringence information about the first sample, and training the classifier based on the birefringence information.
33. The method of claim 32 , wherein the birefringence information comprises one or more birefringence images of the first sample.
34. The method of claim 1 , further comprising providing position information corresponding to the first subset of regions of the first sample to the classifier to train the classifier.
35. The method of claim 34 , wherein the position information comprises an image of the first sample that does not correspond to spectral contributions from the first stain.
36. The method of claim 35 , wherein the position information comprises texture information derived from the image of the first sample.
37. The method of claim 36 , wherein the second stain comprises a counterstain.
38. The method of claim 1 , wherein obtaining the image comprises exposing the first sample to incident radiation and detecting emitted radiation from the first sample, the emitted radiation corresponding to fluorescence emission from the first sample, reflected incident radiation from the first sample, or transmitted incident radiation from the first sample.
39. The method of claim 1 , further comprising using the trained classifier to identify the regions of the second sample.
40. The method of claim 39 , wherein using the trained classifier comprises using the trained classifier to identify the regions of the second sample when the second sample is not stained with the first stain.
41. The method of claim 1 , wherein training the classifier comprises training the classifier to identify regions of the second sample when the second sample is not stained with the first stain.
42. A method, comprising:
applying a stain to a first sample having a plurality of regions, wherein the stain selectively binds to a first set of regions of the first sample;
obtaining an image of the first sample, and analyzing the image to obtain a component image corresponding substantially only to spectral contributions from the stain;
obtaining a birefringence image of the first sample; and
training a classifier to identify regions of a second sample based on information derived from the component image and the birefringence image.
43. The method of claim 42 , wherein the first set of regions is a subset of the regions of the first sample.
44. The method of claim 42 , wherein the first set of regions corresponds to substantially all of the first sample.
45. The method of claim 42 , wherein the birefringence image comprises a set of regions corresponding to a second subset of the regions of the first sample, the second subset of regions of the first sample having optical retardance values that are different from optical retardance values of other regions of the first sample.
46. The method of claim 45 , wherein the second subset of regions and the first set of regions have regions in common.
47. The method of claim 45 , wherein the first set of regions comprises at least some regions that are each adjacent to a region in the second subset of regions.
48. The method of claim 45 , wherein the second subset of regions corresponds to membranes in the first sample.
49. The method of claim 45 , wherein the second subset of regions corresponds to nuclei in the first sample.
50. The method of claim 45 , wherein the information derived from the birefringence image comprises position information corresponding to the second subset of regions.
51. The method of claim 45 , wherein the second subset of regions comprises one or more types of tissue.
52. The method of claim 45 , wherein the second subset of regions comprises one or more types of cells.
53. The method of claim 45 , wherein the second subset of regions comprises one or more types of sub-cellular components.
54. The method of claim 42 , wherein the stain comprises a counterstain.
55. The method of claim 42 , wherein analyzing the image comprises unmixing the image to obtain the component image.
56. The method of claim 42 , wherein analyzing the image comprises separating image color channels to obtain the component image.
57. The method of claim 42 , further comprising:
applying a second stain to the first sample, wherein the second stain binds to a third set of regions of the first sample;
analyzing the image to obtain a second component image corresponding to the second stain; and
training the classifier based on information derived from the second component image.
58. The method of claim 42 , further comprising using the trained classifier to identify the regions of the second sample.
59. The method of claim 58 , wherein using the trained classifier comprises using the trained classifier to identify the regions of the second sample when the second sample is not stained with the stain applied to the first sample.
60. The method of claim 42 , wherein training the classifier comprises training the classifier to identify regions of the second sample when the second sample is not stained with the stain applied to the first sample.
61. The method of claim 42 , wherein the stain comprises at least one member of the group consisting of immunohistochemical agents, cytokeratin, and cadherin.
62. A method, comprising:
applying a stain to a first sample having a plurality of regions, wherein the stain selectively binds to only a first subset of the regions of the first sample;
obtaining an image of the first sample, and analyzing the image to obtain a first component image corresponding substantially only to spectral contributions from the first stain, and a second component image corresponding substantially only autofluorescence contributions from the sample; and
training a classifier to identify regions of a second sample based on information derived from the first and second component images, the identified regions corresponding to the first subset of regions of the first sample.
63. The method of claim 62 , further comprising applying a second stain to the first sample, wherein the second stain selectively binds to a second subset of the regions of the first sample.
64. The method of claim 63 , further comprising:
analyzing the image to obtain a third component image, wherein the third component image corresponds substantially only to spectral contributions from the second stain; and
training the classifier based on information derived from the third component image.
65. The method of claim 62 , wherein the information derived from the first component image comprises position information corresponding to the first subset of regions.
66. The method of claim 62 , wherein the information derived from the first component image comprises information about an amount of the first stain in the first subset of regions.
67. The method of claim 62 , further comprising using the trained classifier to identify the regions of the second sample.
68. The method of claim 67 , wherein using the trained classifier comprises using the trained classifier to identify the regions of the second sample when the second sample is not stained with the stain applied to the first sample.
69. The method of claim 62 , wherein training the classifier comprises training the classifier to identify regions of the second sample when the second sample is not stained with the stain applied to the first sample.
70. The method of claim 62 , wherein the stain comprises at least one member of the group consisting of immunohistochemical agents, cytokeratin, and cadherin.