Diagnostic and prognostic histopathology system using morphometric indices
Determining at least one of a prognosis or a therapy for a patient based on a stained tissue section of the patient. An image of a stained tissue section of a patient is processed by a processing device. A set of features values for a set of cell-based features is extracted from the processed image, and the processed image is associated with a particular cluster of a plurality of clusters based on the set of feature values, where the plurality of clusters is defined with respect to a feature space corresponding to the set of features.
1. A method comprising:
processing, by a processing device, an image of a stained tissue section of a patient;
extracting from the processed image a set of feature values for a set of cell-based features;
associating, based on the set of feature values, the processed image with a particular cluster of a plurality of clusters, wherein the plurality of clusters is defined with respect to a feature space corresponding to the set of features, wherein the particular cluster identifies a morphometric subtype; and
determining whether the morphometric subtype is prognostic-predictive and, in response:
characterizing heterogeneity of a tumor of the patient and evaluating whether heterogeneity is more virulent in terms of prognosis or prediction, and
inferring molecular correlates of a compute morphometric subtype through association with genome-wide molecular data for the tumor.
2. The method of claim 1 wherein the processing of the image comprises normalizing the image in a color map space with respect to a plurality of reference images, and wherein the method further comprises determining at least one of a prognosis or a therapy for the patient based on the particular cluster and historical data associated with the particular cluster.
3. The method of claim 2 wherein the reference images are represented in the feature space as a Gaussian Mixture Model.
4. The method of claim 1 wherein the processing of the image comprises using an image-based model to remove technical variation and biological hetereogeneity.
5. The method of claim 1 wherein the processing of the image comprises applying a graph cut to the subimage to identify nuclei.
6. The method of claim 1 further comprising administering the therapy.
7. The method of claim 1 wherein the processing of the image comprises applying one or more geometric constraints to infer edges between nuclei.
8. A non-transitory computer readable storage medium including instructions that, when executed by a processing device, cause the processing device to perform operations comprising:
determining, by the processing device and based on historical data, whether a morphometric subtype of a set of histology sections is prognostic-predictive; and
in response to determining that the morphometric subtype is prognostic-predictive:
determining, by the processing device and based on the historical data, at least one of a prognosis or a therapy for the morphometric subtype, and
populating, by the processing device, a database with a record that associates the morphometric subtype with the at least one of the prognosis or the therapy.
9. The non-transitory computer readable storage medium of claim 8 wherein the morphometric subtype is determined via clustering of a plurality of feature sets in a feature space, and wherein each of the plurality of feature sets is obtained by processing an image of a respective set of histology sections.
10. The non-transitory computer readable storage medium of claim 9 wherein the processing of the image comprises normalizing the image in a color map space with respect to a plurality of reference images.
11. The non-transitory computer readable storage medium of claim 9 wherein the processing of the image comprises representing the image in the feature space as a Gaussian Mixture Model.
12. The non-transitory computer readable storage medium of claim 9 , further comprising in response to determining that the morphometric subtype is prognostic-predictive:
characterizing tumor heterogeneity and evaluating whether heterogeneity is more virulent in terms of prognosis or prediction, and
inferring molecular correlates of a compute morphometric subtype through association with genome-wide molecular data for a tumor.
13. The non-transitory computer readable storage medium of claim 9 wherein the processing of the image comprises applying a graph cut to the subimage to identify nuclei.
14. The non-transitory computer readable storage medium of claim 9 wherein the processing of the image comprises detecting points of maximum curvature along a contour of a nuclear mask and triangulating the points of maximum curvature.
15. The non-transitory computer readable storage medium of claim 9 wherein the processing of the image comprises applying one or more geometric constraints to infer edges between nuclei.
16. A system comprising:
a memory to store an image of a stained tissue section of a patient; and
a processing device operatively coupled with the memory, the processing device to:
process the image,
extract from the processed image a set of feature values for a set of cell-based features,
associate, based on the set of feature values, the processed image with a particular cluster of a plurality of clusters, wherein the plurality of clusters is defined with respect to a feature space corresponding to the set of features, and
determine a quantification of a tumor property that identifies an intrinsic subtype.
17. The system of claim 16 wherein to process the subimage, the processing device is to normalize the subimage in a color map space with respect to a plurality of reference images.
18. The system of claim 16 wherein the set of cell-based features comprises average nucleus size.
19. The system of claim 16 wherein to process the subimage, the processing device is to apply at least one of a local fitness term or a global fitness term.
20. The system of claim 16 wherein to process the subimage, the processing device is to apply a graph cut to the subimage to identify nuclei.