Systems and methods for processing electronic images with preanalytic adjustment

Patent metadata

Field	Value
Publication number	US-12412248-B2
Application number	US-202217954490-A
Country	US
Kind code	B2
Filing date	Sep 28, 2022
Priority date	Oct 12, 2021
Publication date	Sep 9, 2025
Grant date	Sep 9, 2025

Abstract

A method for processing electronic medical images may include receiving an initial whole slide image of a pathology specimen, receiving information about slide quality aspects to modify, and generating a synthetic whole slide image by applying a machine learning model to modify the received initial whole slide image according to the received information. The pathology specimen may be associated with a patient. The synthetic whole slide image may have a reduced quality as compared to the initial whole slide image.

First claim

What is claimed is:

1. A computer-implemented method for processing electronic medical images, comprising:
receiving an initial whole slide image of a pathology specimen, the pathology specimen being associated with a patient;
receiving information about slide quality aspects to modify; and
generating a synthetic whole slide image by applying a machine learning model to modify the received initial whole slide image according to the received information, wherein the synthetic whole slide image has a reduced quality as compared to the initial whole slide image.

2. The method of claim 1 , wherein receiving information about slide quality aspects to modify includes receiving information about a desired artifact, hair artifact, crack, bubble, dust, dirt, ink scanline, fold, scratch, slice thickness, stain intensity, slide clarity, fixation quality, and/or slide color.

3. The method of claim 1 , wherein receiving an initial whole slide image includes receiving a plurality of initial whole slide images, and wherein generating the synthetic whole slide image includes generating a plurality of synthetic whole slide images, and wherein the method further comprises training a diagnostic machine learning system using the plurality of synthetic whole slide images.

4. The method of claim 1 , wherein receiving information about slide quality aspects to modify includes receiving at least one target whole slide image, the target whole slide image having at least one predetermined defect.

5. The method of claim 4 , wherein generating the synthetic whole slide image by applying the machine learning model includes using a pre-trained neural network to apply neural style transfer to transform the initial whole slide image into the generated whole slide image such that the generated whole slide image includes the predetermined defect.

6. The method of claim 4 , wherein receiving information about slide quality aspects to modify includes receiving at least one pixel-wise annotation with the at least one target whole slide image, wherein the pixel-wise annotation is indicative of an artifact location in the target whole slide image.

7. The method of claim 6 , wherein generating the synthetic whole slide image by applying the machine learning model includes segmenting an artifact and introducing the segmented artifact into the initial whole slide image.

8. The method of claim 7 , further comprising modifying the segmented artifact.

9. The method of claim 7 , further comprising determining one or more locations of the initial whole slide image to introduce the artifact.

10. The method of claim 1 , wherein receiving information about slide quality aspects to modify includes receiving at least one target variable indicative of a predetermined defect.

11. The method of claim 10 , wherein generating the synthetic whole slide image by applying the machine learning model includes using a pre-trained neural network to apply conditional image augmentation to transform the initial whole slide image into the generated whole slide image such that the generated whole slide image includes the predetermined defect.

12. The method of claim 1 , further comprising:
determining at least one salient diagnostic area of interest on the received initial whole slide image; and
determining whether the generated synthetic whole slide image has a sufficient quality for use in a primary diagnosis.

13. The method of claim 1 , further comprising categorizing the generated synthetic whole slide image by tissue type, stain type, diagnosis, quality, or a type of defect or artifact introduced.

14. The method of claim 1 , wherein receiving information about slide quality aspects to modify includes receiving information about a site's specimen transfer and/or slide preparation protocol.

15. The method of claim 1 , wherein receiving information about slide quality aspects to modify includes receiving information about a condition, and wherein generating the synthetic whole slide image by applying the machine learning model includes repeatedly modifying the initial whole slide image until the condition is satisfied.

16. The method of claim 15 , further comprising determining a measure of generalization of the received initial whole slide image based on an extent of modifications performed to satisfy the condition.

17. The method of claim 1 , further comprising:
running a diagnostic system on the generated synthetic whole slide image to determine a diagnosis or a salient diagnostic area of interest; and
determining one or more performance characteristics of the diagnostic system based on a determination by the diagnostic system.

18. The method of claim 1 , further comprising outputting the generated whole slide image to electronic storage and/or a display.

19. 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 an initial whole slide image of a pathology specimen, the pathology specimen being associated with a patient;
receiving information about slide quality aspects to modify; and
generating a synthetic whole slide image by applying a machine learning model to modify the received initial whole slide image according to the received information, wherein the synthetic whole slide image has a reduced quality as compared to the initial whole slide image.

20. A non-transitory computer-readable medium storing instructions that, when executed by a processor, perform operations processing electronic medical images, the operations comprising:
receiving an initial whole slide image of a pathology specimen, the pathology specimen being associated with a patient;
receiving information about slide quality aspects to modify; and
generating a synthetic whole slide image by applying a machine learning model to modify the received initial whole slide image according to the received information, wherein the synthetic whole slide image has a reduced quality as compared to the initial whole slide image.

Assignees

• Paige Ai Inc

Inventors

• Sue Jillian
• Lee Matthew
• Kanan Christopher

Classifications

• G06T2207/30168

  Image quality inspection · CPC title

• G06T2207/30004

  Biomedical image processing · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T7/0012

  Biomedical image inspection · CPC title