Systems and methods for using image processing to generate inferences of biomarker for immunotherapy
Methods and systems are disclosed for image process biological samples to infer biomarker values. An image of a biological sample can be accessed. The image can be segmented into a set of patches. Edge detection performed on each patch can be used to identify one or more biological features represented by the patch. An indication of the one or more features of each patch may be used to generate one or more image level metrics. A value of a biomarker may be inferred using the one or more image level metrics. The value of the biomarker can then be output.
1. A method comprising:
detecting one or more portions of an image that correspond to a biological sample, the image of at least a part of the biological sample;
identifying, for each portion of the one or more portions of the image, a biological feature that is represented by the portion by:
detecting one or more edges based on differences in pixel intensities between adjacent pixels; and
identifying the biological feature based on the detected edges;
generating one or more image level metrics based on the identified biological features;
assigning, based on a first biomarker of interest, a weight to each identified biological feature;
determining a value of a second biomarker represented in the biological sample based on the one or more image level metrics and the weights assigned to each identified biological feature; and
outputting the value of the second biomarker.
2. The method of claim 1 , wherein the first biomarker and the second biomarker are the same biomarker.
3. The method of claim 1 , wherein the first or second biomarker is associated with cancer.
4. The method of claim 3 , wherein the cancer includes lung cancer.
5. The method of claim 3 , wherein the cancer includes a solid tumor.
6. The method of claim 1 , wherein the first or second biomarker is one or more of tumor mutational burden (TMB), programmed death ligand-1 (PD-L1), or interferon gamma (IFNγ) gene signature.
7. The method of claim 1 , wherein the biological feature includes a tumor architecture or nuclear morphology.
8. The method of claim 1 , wherein the one or more image level metrics include a count of a feature, deriving spatial relationships within and between two or more biological features, and/or applying a kernel to identify feature topologies.
9. The method of claim 1 , further comprising generating a prediction of a clinical outcome of a patient based on the value of the second biomarker.
10. The method of claim 1 , further comprising:
detecting a depiction of one or more immune cell sub-populations, wherein the immune cell sub-populations include one or more of small lymphocytes, macrophage, natural killer cells, neutrophils and/or eosinophils; and
determining a quantity and/or a type associated with the one or more immune cell sub-populations, wherein determining the value of the biomarker is further based on the quantity and/or the type associated with one or more immune cell sub-populations.
11. The method of claim 1 , further comprising detecting a nuclear structure of one or more cells depicted in a portion of the image of the one or more portions of the image, wherein determining the value of the biomarker is further based on the nuclear structure of the one or more cells depicted in the image.
12. A system comprising:
a non-transitory computer-readable medium; and
one or more processors communicatively coupled to the non-transitory computer-readable medium, the one or more processors configured to execute processor-executable instructions stored in the non-transitory computer-readable medium to:
detect one or more portions of an image that correspond to a biological sample, the image of at least a part of the biological sample;
identify, for each portion of the one or more portions of the image, a biological feature that is represented by the portion by:
detecting one or more edges based on differences in pixel intensities between adjacent pixels; and
identifying the biological feature based on the detected edges;
generate one or more image level metrics based on the identified biological features;
assign, based on a first biomarker of interest, a weight to each identified biological feature;
determine a value of a second biomarker represented in the biological sample based on the one or more image level metrics and the weights assigned to each identified biological feature; and
output the value of the second biomarker.
13. The system of claim 12 , wherein the first biomarker and the second biomarker are the same biomarker.
14. The system of claim 12 , wherein the first or second biomarker is associated with cancer or is one or more of a tumor mutational burden (TMB), programmed death ligand-1 (PD-L1), or interferon gamma (IFNγ) gene signature.
15. The system of claim 12 , wherein the biological feature includes a tumor architecture or nuclear morphology.
16. The system of claim 12 , wherein the one or more image level metrics include a count of a feature, deriving spatial relationships within and between two or more biological features, and/or applying a kernel to identify feature topologies.
17. The system of claim 12 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to generate a prediction of a clinical outcome of a patient based on the value of the second biomarker.
18. The system of claim 12 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to:
detect a depiction of one or more immune cell sub-populations, wherein the immune cell sub-populations include one or more of small lymphocytes, macrophage, natural killer cells, neutrophils and/or eosinophils; and
determine a quantity and/or a type associated with the one or more immune cell sub-populations, wherein determining the value of the biomarker is further based on the quantity and/or the type associated with one or more immune cell sub-populations.
19. The system of claim 12 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to detect a nuclear structure of one or more cells depicted in a portion of the image of the one or more portions of the image, wherein determining the value of the biomarker is further based on the nuclear structure of the one or more cells depicted in the image.
20. A non-transitory computer-readable medium comprising processor-executable instructions configured to cause one or more processors to:
detect one or more portions of an image that correspond to a biological sample, the image of at least a part of the biological sample;
identify, for each portion of the one or more portions of the image, a biological feature that is represented by the portion by:
detecting one or more edges based on differences in pixel intensities between adjacent pixels; and
identifying the biological feature based on the detected edges;
generate one or more image level metrics based on the identified biological features;
assign, based on a first biomarker of interest, a weight to each identified biological feature;
determine a value of a second biomarker represented in the biological sample based on the one or more image level metrics and the weights assigned to each identified biological feature; and
output the value of the second biomarker.