Batch effect mitigation in digitized images

1. A non-transitory computer-readable medium storing computer-executable instructions that, when executed, cause a processor to perform operations, comprising:

extracting one or more image characterization metrics from respective ones of a plurality of digitized images within an imaging data set, wherein the plurality of digitized images have batch effects;

identifying a plurality of batch effect groups of the digitized images using the one or more image characterization metrics;

partitioning the digitized images within the plurality of batch effect groups into a training set and a validation set, wherein the training set and the validation set respectively comprise some of the plurality of digitized images associated with respective ones of the plurality of batch effect groups;

utilizing the training set to train a machine learning classifier for digital pathology applications; and

operating the machine learning classifier upon the validation set after using the training set to train the machine learning classifier.

2. The non-transitory computer-readable medium of claim 1 ,

wherein the plurality of batch effect groups comprise a first batch effect group and a second batch effect group;

separating a first plurality of digitized images associated with the first batch effect group between the training set and the validation set; and

separating a second plurality of digitized images associated with the second batch effect group between the training set and the validation set.

3. The non-transitory computer-readable medium of claim 1 , wherein the operations further comprise:

providing an additional digitized image to the machine learning classifier after training the machine learning classifier.

4. The non-transitory computer-readable medium of claim 1 , wherein the imaging data set comprises image sets from a plurality of different sites.

5. The non-transitory computer-readable medium of claim 1 , wherein the one or more image characterization metrics comprise a plurality of quality control metrics.

6. The non-transitory computer-readable medium of claim 5 , wherein the plurality of quality control metrics comprise one or more of a tissue color, a background color, a brightness, a contrast, microns per pixel, or magnification.

7. The non-transitory computer-readable medium of claim 5 , wherein the one or more image characterization metrics further comprise information related to biological or tissue properties.

8. The non-transitory computer-readable medium of claim 1 , wherein the plurality of digitized images comprise digitized pathology slides.

9. The non-transitory computer-readable medium of claim 1 , wherein the operations further comprise:

mapping the one or more image characterization metrics associated with the respective ones of the plurality of digitized images into a two-dimensional representation;

forming a two-dimensional cluster plot comprising a plurality of points corresponding to two-dimensional representations of the plurality of digitized images; and

identifying the plurality of batch effect groups within the two-dimensional cluster plot, wherein the plurality of batch effect groups comprise points that are spatially close to one another.

10. A method for mitigating batch effects in a dataset, comprising:

extracting a plurality of quality control metrics from respective ones of a plurality of digitized images within an imaging data set;

mapping the plurality of quality control metrics associated with a respective digitized image of the plurality of digitized images into a two-dimensional representation of the respective digitized image;

plotting a plurality of two-dimensional representations of the plurality of digitized images as points on a two-dimensional cluster plot;

identifying a plurality of batch effect groups within the two-dimensional cluster plot, wherein the plurality of batch effect groups comprise clusters of the points that are spatially close to one another;

partitioning digitized images associated with the plurality of batch effect groups between a training set and a validation set, wherein the training set and the validation set both comprise some of the digitized images associated with respective ones of the plurality of batch effect groups; and

utilizing the training set and the validation set to train a machine learning classifier for digital pathology applications.

11. The method of claim 10 , wherein the plurality of batch effect groups are identified using unsupervised clustering.

12. The method of claim 10 , further comprising:

generating a batch effect summary contact sheet comprising one digitized image associated with respective ones of the plurality of batch effect groups.

13. The method of claim 10 , further comprising:

generating a plurality of batch effect group contact sheets, wherein the plurality of batch effect group contact sheets respectively comprise all digitized images associated with one of the plurality of batch effect groups.

14. The method of claim 10 , further comprising:

forming an output file that comprises a list of image files relating to the plurality of digitized images within the imaging data set, wherein the output file denotes a membership in either the training set or the validation set for each of the image files.

15. The method of claim 10 , wherein the plurality of digitized images comprise digitized images of slides formed from pathology samples.

16. A method, comprising:

extracting one or more image characterization metrics from digitized images having batch effects;

identifying a first group of the digitized images having a first batch effect and a second group of the digitized images having a second batch effect using the one or more image characterization metrics;

separating the first group of the digitized images into a training set and a validation set;

separating the second group of the digitized images into the training set and the validation set; and

operating a machine learning classifier upon one or more of the training set and the validation set to train the machine learning classifier.

17. The method of claim 16 , wherein the one or more image characterization metrics comprise one or more of a tissue color, a background color, a brightness, a contrast, microns per pixel, and a magnification.

18. The method of claim 16 , further comprising:

forming a cluster plot comprising a plurality of points corresponding to the one or more image characterization metrics associated with respective ones of the digitized images; and

identifying the first group of the digitized images and the second group of the digitized images using the cluster plot, wherein the first group of the digitized images and the second group of the digitized images are respectively represented by points that are spatially close to one another.

19. The method of claim 16 , further comprising:

mapping the one or more image characterization metrics associated with respective ones of the digitized images into a two-dimensional representation;

forming a two-dimensional cluster plot comprising a plurality of points corresponding to two-dimensional representations of the digitized images; and

identifying the first group of the digitized images and the second group of the digitized images using the two-dimensional cluster plot, wherein the first group of the digitized images and the second group of the digitized images are respectively represented by points that are spatially close to one another.

20. The method of claim 16 , wherein the digitized images include digitized pathology images or radiology images.