Systems and methods for processing electronic images to provide localized semantic analysis of whole slide images

1. A computer-implemented method for identifying formerly conjoined pieces of tissue in a specimen, the method comprising:

receiving one or more digital images associated with a pathology specimen;

identifying a plurality of pieces of tissue by applying an instance segmentation system to the one or more digital images, the instance segmentation system having been generated by processing a plurality of training images;

determining, using the instance segmentation system, a prediction of whether any of the plurality of pieces of tissue were formerly conjoined in a same tissue sample of the pathology specimen and whether the plurality of pieces of tissue belonged to a different level of the same tissue sample; and

outputting at least one instance segmentation to a digital storage device and/or display, the instance segmentation comprising an indication of whether any of the plurality of pieces of tissue were formerly conjoined.

2. The computer-implemented method of claim 1 , wherein processing the plurality of training images comprises:

receiving one or more digital images of a pathology specimen and associated annotations;

generating a collection of synthetic images and associated annotations based upon the one or more digital images; and

training the instance segmentation system using the collection of synthetic images and associated annotations to perform instance segmentation.

3. The computer-implemented method of claim 2 , wherein generating a collection of synthetic images and associated annotations comprises:

extracting multiple tissue instances from one or more annotated slides and storing the multiple tissue instances separately;

generating a random slide background;

drawing at least one instance from the extracted multiple tissue instances and performing various transformations on the at least one instance; and

placing the at least one instance on the random slide background.

4. The computer-implemented method of claim 3 , wherein storing the multiple tissue instances comprises storing the multiple tissue instances as a Red Blue Green (RBG) image with black pixels for background and/or as a set of polygon coordinates.

5. The computer-implemented method of claim 3 , wherein generating a random slide background comprises:

selecting a random resolution, a number of cores, and/or a noise distribution; and

creating a random slide background from the selection.

6. The computer-implemented method of claim 5 , wherein a noise distribution comprises Gaussian noise, salt-and-pepper noise, and/or periodic noise.

7. The computer-implemented method of claim 3 , wherein performing various transformations on the at least one instance comprises a rotation, a scaling, and/or a change in brightness.

8. The computer-implemented method of claim 3 , wherein placing the at least one instance on the random slide background comprises random placement or pseudo-random placement, wherein pseudo-random placement comprises using heuristics to reproduce common patterns or avoid certain configurations.

9. The computer-implemented method of claim 2 , wherein associated annotations comprise a pixel mask and/or a set of polygons.

10. The computer-implemented method of claim 2 , wherein the instance segmentation system comprises one of a Mask region convolutional neural network (R-CNN), a Deep Mask, a PolyTransform, and/or a Detection Transformer (DETR).

11. A system for identifying formerly conjoined pieces of tissue in a specimen, comprising:

at least one memory storing instructions; and

at least one processor configured to execute the instructions to perform operations comprising:

receiving one or more digital images associated with a pathology specimen;

identifying a plurality of pieces of tissue by applying an instance segmentation system to the one or more digital images, the instance segmentation system having been generated by processing a plurality of training images;

determining, using the instance segmentation system, a prediction of whether any of the plurality of pieces of tissue were formerly conjoined in a same tissue sample of the pathology specimen and whether the plurality of pieces of tissue belonged to a different level of the same tissue sample; and

outputting at least one instance segmentation to a digital storage device and/or display, the instance segmentation comprising an indication of whether any of the plurality of pieces of tissue were formerly conjoined.

12. The system of claim 11 , wherein processing the plurality of training images comprises:

receiving one or more digital images of a pathology specimen and associated annotations;

generating a collection of synthetic images and associated annotations based upon the one or more digital images; and

training the instance segmentation system using the collection of synthetic images and associated annotations to perform instance segmentation.

13. The system of claim 12 , wherein generating a collection of synthetic images and associated annotations comprises:

extracting multiple tissue instances from one or more annotated slides and storing the multiple tissue instances separately;

generating a random slide background;

drawing at least one instance from the extracted multiple tissue instances and performing various transformations on the at least one instance; and

placing the at least one instance on the random slide background.

14. The system of claim 13 , wherein storing the multiple tissue instances comprises storing the multiple tissue instances as a Red Blue Green (RBG) image with black pixels for background and/or as a set of polygon coordinates.

15. The system of claim 13 , wherein generating a random slide background comprises:

selecting a random resolution, a number of cores, and/or a noise distribution; and

creating a random slide background from the selection.

16. The system of claim 15 , wherein a noise distribution comprises Gaussian noise, salt-and-pepper noise, and/or periodic noise.

17. The system of claim 13 , wherein performing various transformations on the at least one instance comprises a rotation, a scaling, and/or a change in brightness.

18. The system of claim 12 , wherein associated annotations comprise a pixel mask and/or a set of polygons.

19. The system of claim 12 , wherein the instance segmentation system comprises one of a Mask region convolutional neural network (R-CNN), a Deep Mask, a PolyTransform, and/or a Detection Transformer (DETR).

20. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform operations for identifying formerly conjoined pieces of tissue in a specimen, the operations comprising:

receiving one or more digital images associated with a pathology specimen;

identifying a plurality of pieces of tissue by applying an instance segmentation system to the one or more digital images, the instance segmentation system having been generated by processing a plurality of training images;

determining, using the instance segmentation system, a prediction of whether any of the plurality of pieces of tissue were formerly conjoined in a same tissue sample of the pathology specimen and whether the plurality of pieces of tissue belonged to a different level of the same tissue sample; and

outputting at least one instance segmentation to a digital storage device and/or display, the instance segmentation comprising an indication of whether any of the plurality of pieces of tissue were formerly conjoined.