Ovarian toxicity assessment in histopathological images using deep learning

1. A method comprising:

obtaining a set of images of tissue slices from one or more ovaries treated with an amount of a compound;

inputting the set of images into a neural network model constructed as a one-stage detector using focal loss as at least a portion of the loss function;

predicting, using the neural network model, coordinates for a bounding box around one or more objects within the set of images that are identified as corpora lutea (CL);

outputting, using the neural network model, the set of images with the bounding box around the one or more objects that are identified as the CL based on the coordinates predicted for the bounding box;

counting the bounding boxes within the set of images output from the neural network model to obtain a CL count for the ovary; and

determining an ovarian toxicity of the compound at the amount based on the CL count for the ovary, wherein the determining the ovarian toxicity comprises examining the CL count for the ovary in relation to a given condition, indicative of whether the compound at the amount is toxic to the ovary.

2. The method of claim 1 , further comprising identifying, using the neural network model, the one or more objects within the set of images as being the CL, wherein the identifying comprises generating a probability score for each object of the one or more objects that is identified as the CL, and the counting comprises only counting the bounding boxes within the set of images around the one or more objects identified as the CL and having a probability score greater than a predetermined probability score threshold.

3. The method of claim 1 , wherein the examining the CL count for the ovary comprises comparing the CL count for the ovary to a predetermined toxicity threshold, when the CL count is above the predetermined toxicity threshold, determining the compound at the amount is not toxic to the ovary, and when the CL count is below or equal to the predetermined toxicity threshold, determining the compound at the amount is toxic to the ovary.

4. The method of claim 1 , wherein the set of images of the tissue slices are obtained from an ovary of a subject treated with the amount of the compound, and the method further comprises:

obtaining another set of images of tissue slices from another ovary of the subject treated with the amount of the compound;

generating, using the neural network model, the another set of images with a bounding box around one or more objects of objects that are identified as the CL within the another set of images based on coordinates predicted for the bounding box; and

counting the bounding boxes within the another set of images output from the neural network model to obtain another CL count for the another ovary,

wherein the examining the CL count for the ovary comprises:

averaging the CL count for the ovary and the another CL count for the another ovary to obtain an averaged CL count,

wherein the ovarian toxicity of the compound at the amount is determined based on the average CL.

5. The method of claim 1 , further comprising:

obtaining another set of images of tissue slices from one or more other ovaries that are either untreated or treated with a different amount of the compound;

generating, using the neural network model, the another set of images with a bounding box around one or more objects of objects that are identified as the CL within the another set of images based on coordinates predicted for the bounding box; and

counting the bounding boxes within the another set of images output from the neural network model to obtain another CL count for the ovary,

wherein the examining the CL count for the ovary comprises determining a trend between the CL count for the one or more ovaries and the another CL count for the one or more other ovaries, and the ovarian toxicity of the compound is determined at the amount or at the different amount based on the trend.

6. The method of claim 1 , further comprising:

obtaining another set of images of tissue slices from one or more other ovaries that are treated with an amount of a different compound;

generating, using the neural network model, the another set of images with a bounding box around one or more objects of objects that are identified as the CL within the another set of images based on coordinates predicted for the bounding box;

counting the bounding boxes within the another set of images output from the neural network model to obtain another CL count for the ovary of the another subject; and

determining an ovarian toxicity of the different compound at the amount based on the CL count for the one or more other ovaries.

7. The method of claim 1 , further comprising:

providing the set of images with the bounding box around each object of the objects that is identified as the CL, the CL count for the one or more ovaries, the ovarian toxicity of the compound at the amount, or any combination thereof; and

administering a treatment with the compound based on the ovarian toxicity of the compound at the amount.

8. A computer-program product tangibly embodied in a non-transitory machine-readable storage medium, including instructions configured to cause one or more data processors to perform actions including:

obtaining a set of images of tissue slices from one or more ovaries treated with an amount of a compound;

inputting the set of images into a neural network model constructed as a one-stage detector using focal loss as at least a portion of the loss function;

predicting, using the neural network model, coordinates for a bounding box around one or more objects within the set of images that are identified as corpora lutea (CL);

outputting, using the neural network model, the set of images with the bounding box around the one or more objects that are identified as the CL based on the coordinates predicted for the bounding box;

counting the bounding boxes within the set of images output from the earning neural network model to obtain a CL count for the ovary; and

determining an ovarian toxicity of the compound at the amount based on the CL count for the ovary, wherein the determining the ovarian toxicity comprises examining the CL count for the ovary in relation to a given condition, indicative of whether the compound at the amount is toxic to the ovary.

9. The computer-program product of claim 8 , wherein the actions further include identifying, by the neural network model, the one or more objects within the set of images as being the CL, and wherein the identifying comprises generating a probability score for each object of the one or more objects that is identified as the CL, and the counting comprises only counting the bounding boxes within the set of images around the one or more objects identified as the CL and having a probability score greater than a predetermined probability score threshold.

10. The computer-program product of claim 8 , wherein the examining the CL count for the ovary comprises comparing the CL count for the ovary to a predetermined toxicity threshold, when the CL count is above the predetermined toxicity threshold, determining the compound at the amount is not toxic to the ovary, and when the CL count is below or equal to the predetermined toxicity threshold, determining the compound at the amount is toxic to the ovary.

11. The computer-program product of claim 8 , wherein the set of images of the tissue slices are obtained from an ovary of a subject treated with the amount of the compound and the actions further include:

obtaining another set of images of tissue slices from another ovary of the subject treated with the amount of the compound;

generating, using the neural network model, the another set of images with a bounding box around one or more objects of objects that are identified as the CL within the another set of images based on coordinates predicted for the bounding box; and

counting the bounding boxes within the another set of images output from the neural network model to obtain another CL count for the another ovary,

wherein the examining the CL count for the ovary comprises:

averaging the CL count for the ovary and the another CL count for the another ovary to obtain an averaged CL count,

wherein the ovarian toxicity of the compound at the amount is determined based on the average CL.

12. The computer-program product of claim 8 , wherein the actions further include:

obtaining another set of images of tissue slices from one or more other ovaries that are either untreated or treated with a different amount of the compound;

generating, using the neural network model, the another set of images with a bounding box around one or more objects of objects that are identified as the CL within the another set of images based on coordinates predicted for the bounding box; and

counting the bounding boxes within the another set of images output from the neural network model to obtain another CL count for the ovary,

wherein the examining the CL count for the ovary comprises determining a trend between the CL count for the one or more ovaries and the another CL count for the one or more other ovaries, and the ovarian toxicity of the compound is determined at the amount or at the different amount based on the trend.

13. The computer-program product of claim 8 , the actions further include:

obtaining another set of images of tissue slices from one or more other ovaries that are treated with an amount of a different compound;

generating, using the neural network model, the another set of images with a bounding box around one or more objects of objects that are identified as the CL within the another set of images based on coordinates predicted for the bounding box;

counting the bounding boxes within the another set of images output from the neural network model to obtain another CL count for the ovary of the another subject; and

determining an ovarian toxicity of the different compound at the amount based on the CL count for the one or more other ovaries.

14. The computer-program product of claim 8 , wherein the actions further include providing the set of images with the bounding box around each object of the objects that is identified as the CL, the CL count for the one or more ovaries, the ovarian toxicity of the compound at the amount, or any combination thereof.

15. A system comprising:

one or more data processors; and

a non-transitory computer readable storage medium containing instructions which, when executed on the one or more data processors, cause the one or more data processors to perform actions including:

obtaining a set of images of tissue slices from one or more ovaries treated with an amount of a compound;

inputting the set of images into a neural network model constructed as a one-stage detector using focal loss as at least a portion of the loss function;

predicting, using the neural network model, coordinates for a bounding box around one or more objects within the set of images that are identified as corpora lutea (CL);

outputting, using the neural network model, the set of images with the bounding box around the one or more objects that are identified as the CL based on the coordinates predicted for the bounding box;

counting the bounding boxes within the set of images output from the neural network model to obtain a CL count for the ovary; and

determining an ovarian toxicity of the compound at the amount based on the CL count for the ovary, wherein the determining the ovarian toxicity comprises examining the CL count for the ovary in relation to a given condition, indicative of whether the compound at the amount is toxic to the ovary.

16. The system of claim 15 , wherein the actions further include identifying, by the neural network model, the one or more objects within the set of images as being the CL, and wherein the identifying comprises generating a probability score for each object of the one or more objects that is identified as the CL, and the counting comprises only counting the bounding boxes within the set of images around the one or more objects identified as the CL and having a probability score greater than a predetermined probability score threshold.

17. The system of claim 15 , wherein the examining the CL count for the ovary comprises comparing the CL count for the ovary to a predetermined toxicity threshold, when the CL count is above the predetermined toxicity threshold, determining the compound at the amount is not toxic to the ovary, and when the CL count is below or equal to the predetermined toxicity threshold, determining the compound at the amount is toxic to the ovary.

18. The system of claim 15 , wherein the set of images of the tissue slices are obtained from an ovary of a subject treated with the amount of the compound, and the actions further include:

obtaining another set of images of tissue slices from another ovary of the subject treated with the amount of the compound;

generating, using the neural network model, the another set of images with a bounding box around one or more objects of objects that are identified as the CL within the another set of images based on coordinates predicted for the bounding box; and

counting the bounding boxes within the another set of images output from the neural network model to obtain another CL count for the another ovary,

wherein the examining the CL count for the ovary comprises:

averaging the CL count for the ovary and the another CL count for the another ovary to obtain an averaged CL count,

wherein the ovarian toxicity of the compound at the amount is determined based on the average CL.

19. The system of claim 15 , wherein the actions further include:

obtaining another set of images of tissue slices from one or more other ovaries that are either untreated or treated with a different amount of the compound;

generating, using the neural network model, the another set of images with a bounding box around one or more objects of objects that are identified as the CL within the another set of images based on coordinates predicted for the bounding box; and

counting the bounding boxes within the another set of images output from the neural network model to obtain another CL count for the ovary,

wherein the examining the CL count for the ovary comprises determining a trend between the CL count for the one or more ovaries and the another CL count for the ovary, and the ovarian toxicity of the compound is determined at the amount or at the different amount based on the trend.

20. The system of claim 15 , the actions further include:

obtaining another set of images of tissue slices from one or more other ovaries that are treated with an amount of a different compound;

generating, using the neural network model, the another set of images with a bounding box around one or more objects of objects that are identified as the CL within the another set of images based on coordinates predicted for the bounding box;

counting the bounding boxes within the another set of images output from the neural network model to obtain another CL count for the ovary of the another subject; and

determining an ovarian toxicity of the different compound at the amount based on the CL count for the one or more other ovaries.