EP-4742195-A2 - SYSTEMS AND METHODS TO PROCESS ELECTRONIC IMAGES TO PROVIDE IMAGE-BASED CELL GROUP TARGETING

Abstract

Systems and methods are disclosed for grouping cells in a slide image that share a similar target, comprising receiving a digital pathology image corresponding to a tissue specimen, applying a trained machine learning system to the digital pathology image, the trained machine learning system being trained to predict at least one target difference across the tissue specimen, determining, using the trained machine learning system, one or more predicted clusters, each of the predicted clusters corresponding to a subportion of the tissue specimen associated with a target, using an artificial intelligence (Al)-predicted segmentation with a clustering heuristic to determine one or more optimal sampling locations of the tissue specimen to maximize information gained about the target difference across the tissue specimen, and providing the predicted clusters and the one or more optimal sampling locations for output to a display.

Inventors

• CEBALLOS LENTINI, Rodrigo
• KANAN, CHRISTOPHER
• DOGDAS, Belma

Assignees

• PAIGE.AI, INC.

Dates

Publication Date

20260513

Application Date

20210802

Claims (13)

1. A computer-implemented method for grouping cells in a slide image that share a similar target, the method comprising: receiving a digital pathology image corresponding to a tissue specimen; applying a trained machine learning system to the digital pathology image, the trained machine learning system being trained to predict at least one target difference across the tissue specimen; determining, using the trained machine learning system, one or more predicted clusters, each of the predicted clusters corresponding to a subportion of the tissue specimen associated with a target; using an artificial intelligence (Al)-predicted segmentation with a clustering heuristic to determine one or more optimal sampling locations of the tissue specimen to maximize information gained about the target difference across the tissue specimen; and providing the predicted clusters and the one or more optimal sampling locations for output to a display.
2. The computer-implemented method of claim 1, further comprising: using available sequencing techniques to determine a target composition of one or more samples at the one or more optimal sampling locations; and using the Al-predicted segmentation to infer a spatial distribution over factors in the tissue specimen based on the target composition.
3. The computer-implemented method of claim 1, wherein the clustering heuristic comprises k-means clustering and/or hierarchical clustering..
4. The computer-implemented method of claim 1, wherein the one or more optimal sampling locations are determined based on genetic targets.
5. The computer-implemented method of claim 1, wherein each of the one or more predicted clusters comprises antigen clusters.
6. The computer-implemented method of claim 1, further comprising: mapping, using the trained machine learning system, the predicted clusters to at least one known biological target; and providing the at least one known biological target for output to the display.
7. The computer-implemented method of claim 1, further comprising: determining, using the predicted clusters and the trained machine learning system, predicted mappings comprising a map from any of the predicted clusters to a predicted mutation; and providing the predicted mutation for output to the display.
8. The computer-implemented method of claim 1, further comprising: determining one or more pixel masks, each pixel mask segmenting the digital pathology image into one or more of the predicted clusters; and providing the pixel masks for output to the display.
9. The computer-implemented method of claim 1, further comprising: using a spatial distribution and spatial relation of the predicted clusters to determine a treatment decision associated with the tissue specimen.
10. The computer-implemented method of claim 1, further comprising: using flow cytometry and/or mass spectrometry techniques to determine physical and/or chemical characteristics of one or more samples from the one or more optimal sampling locations.
11. The computer-implemented method of claim 1, further comprising: predicting relative proportions of the predicted clusters across cancer cells within the digital pathology image.
12. A system for grouping cells in a slide image that share a similar target, comprising: at least one memory storing instructions; and at least one processor configured to execute the instructions to perform a method according to any one of claims 1 to 11.
13. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform a method according to any one of claims 1 to 11.

Description

RELATED APPLICATION(S) This application claims priority to U.S. Provisional Application No. 63/061,056 filed August 4, 2020, the entire disclosure of which is hereby incorporated herein by reference in its entirety. FIELD OF THE DISCLOSURE Various embodiments of the present disclosure pertain generally to image-based cell group targeting and related image processing methods. More specifically, particular embodiments of the present disclosure relate to systems and methods for cell group targeting based on processing images of tissue specimens. BACKGROUND As personalized and targeted disease treatment options become viable, a need for fast, affordable and scalable genetic sequencing of diseases grows. Currently, one approach to this issue is to sequence large parts of tumor tissue to find usable genetic or epigenetic targets. However, the heterogeneity of most cancers means that the most common targets across all tumors in a patient are not necessarily the most effective targets. Also, this approach does not take into consideration that there may be healthy cells in the extracted tumor which may make it more difficult to target the cancer with minimal side effects. One solution to this problem may be to sequence smaller areas of the tumor to better understand the different mutations present across the cancer cell population in a patient, and how they differentiate from the healthy cells surrounding it. Techniques in spatial transcriptomics that may do this at a high resolution remain unproven, expensive, not scalable, and destroy the tissue. One or more embodiments of the present disclosure may overcome the above-described problems. The background description provided herein is for the purpose of generally presenting the context of the disclosure. Unless otherwise indicated herein, the materials described in this section are not prior art to the claims in this application and are not admitted to be prior art, or suggestions of the prior art, by inclusion in this section. SUMMARY According to certain aspects of the present disclosure, systems and methods are disclosed for grouping cells in a slide image that share a similar target. A method for grouping cells in a slide image that share a similar target, comprises receiving a digital pathology image corresponding to a tissue specimen, applying a trained machine learning system to the digital pathology image, the trained machine learning system being trained to predict at least one target difference across the tissue specimen, and determining, using the trained machine learning system, one or more predicted clusters, each of the predicted clusters corresponding to a subportion of the tissue specimen associated with a target. A system for grouping cells in a slide image that share a similar target, comprising at least one memory storing instructions; and at least one processor configured to execute the instructions to perform operations comprising receiving a digital pathology image corresponding to a tissue specimen, applying a trained machine learning system to the digital pathology image, the trained machine learning system being trained to predict at least one target difference across the tissue specimen, and determining, using the trained machine learning system, one or more predicted clusters, each of the predicted clusters corresponding to a subportion of the tissue specimen associated with a target. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform operations for grouping cells in a slide image that share a similar target, comprises receiving a digital pathology image corresponding to a tissue specimen, applying a trained machine learning system to the digital pathology image, the trained machine learning system being trained to predict at least one target difference across the tissue specimen, and determining, using the trained machine learning system, one or more predicted clusters, each of the predicted clusters corresponding to a subportion of the tissue specimen associated with a target. It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosed embodiments, as claimed. BRIEF DESCRIPTION OF THE DRAWINGS The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate various exemplary embodiments and together with the description, serve to explain the principles of the disclosed embodiments. FIG. 1A illustrates an exemplary block diagram of a system and network for grouping cells in a whole slide image (WSI) that share similar targets, according to an exemplary embodiment of the present disclosure.FIG. 1B illustrates an exemplary block diagram of the disease detection platform 100, according to an exemplary embodiment of the present disclosure.FIG. 2 is a flowchart of an exemplary method for grouping cells