Pathology predictions on unstained tissue
A method for training a pattern recognizer to identify regions of interest in unstained images of tissue samples is provided. Pairs of images of tissue samples are obtained, each pair including an unstained image of a given tissue sample and a stained image of the given tissue sample. An annotation (e.g., drawing operation) is then performed on the stained image to indicate a region of interest. The annotation information, in the form of a mask surrounding the region of interest, is then applied to the corresponding unstained image. The unstained image and mask are then supplied to train a pattern recognizer. The trained pattern recognizer can then be used to identify regions of interest within novel unstained images.
1. A method of training a pattern recognizer to identify regions of interest in a tissue sample, comprising the steps of:
obtaining magnified digital images of the tissue sample, one of which represents the tissue sample after having been stained with a staining agent (“stained image”) and one of which represents the tissue sample in an unstained state (“unstained image”),
annotating the stained image so as to form a mask surrounding a region of interest (“ROI”) in the stained image;
transferring the mask from the stained image to the unstained image; and
training the pattern recognizer to identify ROIs in unstained images of tissue samples using the unstained image having the transferred mask as a training example.
2. The method of claim 1 , further comprising aligning the stained and unstained images.
3. The method of claim 1 , wherein the pattern recognizer comprises an artificial neural network.
4. The method of claim 3 , wherein the artificial neural network is a convolutional neural network.
5. The method of claim 1 , wherein the mask surrounds at least one closed polygonal region in the stained image.
6. The method of claim 1 , wherein the tissue sample is a sample of lymph tissue.
7. The method of claim 1 , wherein the tissue sample is a sample of breast tissue.
8. The method of claim 1 , wherein the tissue sample is a sample of prostate tissue.
9. The method of claim 1 , wherein the staining agent comprises an immunohistochemical (IHC) stain.
10. The method of claim 1 , wherein the staining agent comprises a haematoxylin and eosin stain.
11. A method of identifying a region of interest in an unstained tissue sample, comprising:
providing a pattern recognizer trained to identify regions of interest in unstained magnified digital images of tissue samples;
obtaining a magnified digital image of an unstained tissue sample; and
processing the magnified digital image of the unstained tissue sample with the pattern recognizer to generate annotation information indicative of a region of interest in the unstained tissue sample.
12. The method of claim 11 , wherein the pattern recognizer comprises an artificial neural network.
13. The method of claim 12 , wherein the artificial neural network is a convolutional neural network.
14. The method of claim 11 , wherein the annotation information is indicative of at least one closed polygonal region in the unstained image.
15. The method of claim 11 , wherein the tissue sample is a sample of lymph tissue.
16. The method of claim 11 , wherein the tissue sample is a sample of breast tissue.
17. The method of claim 11 , wherein the tissue sample is a sample of prostate tissue.
18. The method of claim 11 , further comprising:
obtaining a magnified digital image of the tissue sample after having been stained with a staining agent (“stained image”);
wherein the pattern recognizer trained to identify regions of interest in unstained magnified digital images of tissue samples is trained to identify regions of interest in unstained magnified digital images of tissue samples based on unstained magnified digital images of tissue samples and magnified digital images of such tissue samples after having been stained with a staining agent, and wherein processing the magnified digital image of the unstained tissue sample with the pattern recognizer to generate annotation information comprises processing the magnified digital image of the unstained tissue sample and the magnified digital image of the tissue sample after having been stained with the staining agent with the pattern recognizer to generate the annotation information.
19. The method of claim 18 , further comprising aligning the unstained image and the magnified digital image of the unstained tissue sample.
20. A system comprising:
a memory storing non-transient processor readable instructions; and
one or more processors arranged to read and execute instructions stored in said memory;
wherein said processor readable instructions comprise instructions arranged to control the system to carry out a method according to claim 11 .