Classifying image features
View Patent ↗Methods are disclosed for classifying different parts of a sample into respective classes based on an image stack that includes one or more images.
1. A method for assessing a biological sample, the method comprising using one or more electronic processors to perform the following steps:
decomposing a set of n spectral images of the sample into a set of m unmixed images, wherein each member of the unmixed image set corresponds to a spectral contribution from a different component in the sample, and wherein n≧4; and
classifying spatial locations in the sample corresponding to pixels in the set of unmixed images based on an image stack comprising one or more members of the set of unmixed images to assess whether each spatial location corresponds to one or more of p classes.
2. The method of claim 1 , wherein the decomposing comprises decomposing the set of n spectral images into a set of m≦3 unmixed images.
3. The method of claim 1 , wherein the decomposing comprises decomposing the set of n spectral images into a set of m<n unmixed images.
4. The method of claim 1 , wherein the set of spectral images comprises n=4 images and the set of unmixed images comprises m=3 images.
5. The method of claim 1 , wherein the classifying comprises assessing whether each spatial location corresponds to one or more of p=m classes.
6. The method of claim 1 , wherein the classifying comprises assessing whether each spatial location corresponds to one or more of p<m classes.
7. The method of claim 1 , wherein p>1.
8. The method of claim 1 , wherein p=1.
9. The method of claim 1 , wherein one of the p classes is a disease state.
10. The method of claim 1 , wherein one of the p classes is a cell or tissue type.
11. The method of claim 1 , wherein one of the p classes is a biological structure type.
12. The method of claim 1 , wherein the biological sample comprises cells.
13. The method of claim 1 , wherein the decomposing further comprises determining amounts of at least some of the sample components in at least some of the spatial locations of the sample.
14. The method of claim 13 , wherein the decomposing further comprises determining amounts of each of the sample components in at least some of the spatial locations of the sample.
15. The method of claim 1 , wherein the classifying comprises classifying at least some of the spatial locations based on information derived from pixels in the image stack neighboring the pixels to which the spatial locations correspond.
16. The method of claim 15 , wherein the classifying comprises classifying the at least some of the spatial locations based on both spectral and spatial information derived from the neighboring pixels.
17. The method of claim 1 , further comprising obtaining the set of spectral images by spectrally filtering illumination light directed to the sample to select light in each of a plurality of different wavelength bands, wherein each member of the set of spectral images corresponds to a different one of the wavelength bands.
18. The method of claim 1 , further comprising obtaining the set of spectral images by spectrally filtering light received from the sample in each of a plurality of different wavelength bands, wherein each member of the set of spectral images corresponds to a different one of the wavelength bands.
19. The method of claim 1 , wherein classifying spatial locations corresponding to pixels in the set of unmixed images comprises:
positioning a sampling window within the image stack to select a portion of the image stack for classification, the selected portion comprising multiple pixels;
classifying the selected portion to assess whether the selected portion corresponds to the one or more of p classes, wherein each of the multiple pixels in the selected portion is provisionally assessed as having the same classification with respect to the one or more of p classes as the selected portion;
translating the sampling window to select a second portion of the image stack comprising multiple pixels for classification and classifying the second portion to assess whether the second portion corresponds to the one or more of p classes, wherein each of the multiple pixels in the second portion is provisionally assessed as having the same classification with respect to the one or more of p classes as the second portion;
repeating the translating and classifying for additional portions of the image stack until at least some of the pixels in the image stack have been provisionally assessed multiple times as part of different portions selected by the sampling window; and
assessing whether each of at least some of the spatial locations corresponding to the pixels that have been provisionally assessed multiple times corresponds to the one or more of p classes based on the multiple provisional assessments.
20. The method of claim 1 , further comprising using a machine learning classifier to classify the spatial locations, wherein the machine learning classifier comprises at least one member selected from the group consisting of a neural network-based classifier, a genetic algorithm-based classifier, and a support vector machine-based classifier.
21. A system for assessing a biological sample, the system comprising:
a light source;
light conditioning optics positioned to direct light from the source to the sample;
light collecting optics positioned to direct light from the sample to a detector;
a detector configured to receive light from the light collecting optics and to record a set of n spectral images of the sample based on the received light, wherein n≧4; and
an electronic processor connected to the detector and configured to:
decompose the set of n spectral images into a set of m unmixed images, wherein each member of the unmixed image set corresponds to a spectral contribution from a different component in the sample; and
classify spatial locations in the sample corresponding to pixels in the set of unmixed images based on an image stack comprising one or more members of the set of unmixed images to assess whether each spatial location corresponds to one or more of p classes.
22. The system of claim 21 , wherein the electronic processor is configured to decompose the set of n spectral images into a set of m≦3 unmixed images.
23. The system of claim 21 , wherein the electronic processor is configured to decompose the set of n spectral images into a set of m<n unmixed images.
24. The system of claim 21 , wherein the detector is configured to record a set of n=4 spectral images of the sample, and wherein the electronic processor is configured to decompose the set of n=4 spectral images into a set of m=3 unmixed images.
25. The system of claim 21 , wherein the electronic processor is configured to assess whether each spatial location corresponds to one or more of p≦m classes.
26. The system of claim 21 , wherein the electronic processor is configured to assess whether each spatial location corresponds to one or more of p>1 classes.
27. The system of claim 26 , wherein one of the p classes is a disease state.
28. The system of claim 26 , wherein one of the p classes is a cell or tissue type.
29. The system of claim 26 , wherein one of the p classes is a biological structure type.
30. The system of claim 21 , wherein the electronic processor is further configured to determine amounts of at least some of the sample components in at least some of the spatial locations of the sample.