Systems and methods for processing electronic images for ranking loss and grading

Patent metadata

Field	Value
Publication number	US-12573218-B2
Application number	US-202318161688-A
Country	US
Kind code	B2
Filing date	Jan 30, 2023
Priority date	Jan 31, 2022
Publication date	Mar 10, 2026
Grant date	Mar 10, 2026

Abstract

A computer-implemented method for processing medical images, the method comprising receiving a plurality of medical images of at least one pathology specimen, the pathology specimen being associated with a patient. The method may further comprise dividing the one or more medical images into a plurality of tiles and predicting, using a machine learning system, proportions of each type of cancer sub-category for the plurality of tiles, the machine learning system having been trained by ranking loss. The method may further include determining an overall grade of cancer for the one or more medical images.

First claim

What is claimed is:

1 . A computer-implemented method for processing electronic medical images, comprising:
receiving one or more medical images of at least one pathology specimen associated with a patient; dividing the one or more medical images into a plurality of tiles; predicting, using a machine learning system, proportions of each type of cancer sub-category for the plurality of tiles, the machine learning system having been trained by ranking loss, the machine learning system having been trained by:
dividing a plurality of electronic images into tiles; and
for each of the plurality of electronic images:
predicting proportions of any of a plurality of cancer sub-categories;
determining a loss for each less than relation based on ground-truth proportions;
determining a total loss based on the loss for each less than relation;
determining a backward pass; and
updating the machine learning system using the total loss and backward pass; and

determining an overall grade of cancer for the one or more medical images.

2 . The method of claim 1 , wherein the overall grade of cancer is output as a Gleason score prediction.

3 . The method of claim 1 , wherein predicting proportions of each type of cancer includes predicting proportions of each cancer sub-category.

4 . The method of claim 3 , wherein during training, the ranking loss is determined for each cancer sub category.

5 . The method of claim 4 , wherein the total ranking loss of the machine learning system is calculated using a ranking loss formula Lrank=x+1 if x>0 & Lrank=ex if x<0.

6 . The method of claim 5 , wherein the machine learning system is trained using weak labels to learn more granular labels using the rank loss formula.

7 . The method of claim 3 , further comprising:
outputting the proportion of each cancer sub-category.

8 . The method of claim 1 , further comprising:
outputting a segmentation map displaying a location of each cancer sub-category.

9 . A system for processing electronic medical images, the system 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 medical images of at least one pathology specimen associated with a patient;
dividing the one or more medical images into a plurality of tiles;
predicting, using a machine learning system, proportions of each type of cancer sub-category for the plurality of tiles, the machine learning system having been trained by ranking loss, the machine learning system having been trained by:
dividing a plurality of electronic images into tiles; and
for each of the plurality of electronic images:
predicting proportions of any of a plurality of cancer sub-categories;
determining a loss for each less than relation based on ground-truth proportions;
determining a total loss based on the loss for each less than relation;
determining a backward pass; and
updating the machine learning system using the total loss and backward pass; and

determining an overall grade of cancer for the one or more medical images.

10 . The system of claim 9 , wherein the overall grade of cancer is output as a Gleason score prediction.

11 . The system of claim 9 , wherein predicting proportions of each type of cancer includes predicting proportions of each cancer sub-category.

12 . The system of claim 11 , wherein during training, the ranking loss is determined for each cancer sub category.

13 . The system of claim 12 , wherein the total ranking loss of the machine learning system is calculated using a ranking loss formula Lrank=x+1 if x>0 & Lrank=ex if x<0.

14 . The system of claim 13 , wherein the machine learning system is trained using weak labels to learn more granular labels using the rank loss formula.

15 . The system of claim 11 , further comprising:
outputting the proportion of each cancer sub-category.

16 . The system of claim 9 , further comprising:
outputting a segmentation map displaying a location of each cancer sub-category.

17 . A non-transitory computer-readable medium storing instructions that, when executed by a processor, perform operations processing electronic digital medical images, the operations comprising:
receiving one or more medical images of at least one pathology specimen associated with a patient; dividing the one or more medical images into a plurality of tiles; predicting, using a machine learning system, proportions of each type of cancer sub-category for the plurality of tiles, the machine learning system having been trained by ranking loss, the machine learning system having been trained by:
dividing a plurality of electronic images into tiles; and
for each of the plurality of electronic images:
predicting proportions of any of a plurality of cancer sub-categories;
determining a loss for each less than relation based on ground-truth proportions;
determining a total loss based on the loss for each less than relation;
determining a backward pass; and
updating the machine learning system using the total loss and backward pass; and

determining an overall grade of cancer for the one or more medical images.

18 . The non-transitory computer-readable medium of claim 17 , wherein the overall grade of cancer is output as a Gleason score prediction.

19 . The non-transitory computer-readable medium of claim 17 , wherein predicting proportions of each type of cancer includes predicting proportions of each cancer sub-category.

20 . The non-transitory computer-readable medium of claim 17 , wherein during training, the ranking loss is determined for each cancer sub category.

Assignees

• Paige Ai Inc

Inventors

• Aghdam Hamed
• Kanan Christopher

Classifications

• G06T2207/30024

  Cell structures in vitro; Tissue sections in vitro · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/20021

  Dividing image into blocks, subimages or windows · CPC title

• G06T7/0012

  Biomedical image inspection · CPC title

• G16H50/30

  for calculating health indices; for individual health risk assessment · CPC title