Few-Shot Learning for Whole Slide Image Tissue Recognition

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods for determining tissue characteristics, such as cancer type and grade, in histopathology images are unreliable, expensive, and time-consuming, often requiring verification by multiple pathologists.

Innovation Solution

The use of computer vision and machine learning, specifically training a neural network to classify digital histopathology images, allows for the automated identification of tissue characteristics by analyzing image patches and utilizing input from pathologists to improve accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional pathologist review methods are used to determine tissue characteristics, then diagnostic accuracy can be maintained through human expertise, but the process becomes time-consuming and expensive requiring multiple pathologists for verification

Engineering Contradiction:

Improvediagnostic accuracyVSAvoidtime consumption

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary automated analysis of tissue images using machine learning models to pre-identify regions of interest and potential diagnostic features before pathologist review. This preliminary processing filters and prioritizes cases, so that pathologists only need to verify critical findings rather than review all images from scratch, significantly reducing their time investment while maintaining diagnostic accuracy.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

An automated machine learning-based image analysis system serves as an intermediary between the tissue sample and the pathologist. This intermediary performs initial diagnostic assessment, generates preliminary reports, and highlights areas requiring human review, thereby reducing the time pathologists spend on routine cases while preserving their expertise for complex or ambiguous diagnoses.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If traditional multiple pathologist verification is used to ensure reliable tissue characterization, then diagnostic reliability is improved, but the cost and complexity of the process increases

Engineering Contradiction:

Improvediagnostic reliabilityVSAvoidprocess complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system replaces the mechanical process of multiple pathologists independently reviewing and verifying each case with an automated machine learning model that consistently applies diagnostic criteria. This substitution eliminates human variability and fatigue while maintaining reliability through the model's trained expertise, reducing both process complexity and resource requirements.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Productivity

If computer vision and machine learning are used to automate tissue characterization, then speed and efficiency are improved, but the requirement for large amounts of training data increases system complexity

Engineering Contradiction:

Improveprocessing speedVSAvoidtraining data requirement

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

Inventive Principle:

Principle #10Preliminary action

4Productivity

If automated machine learning classification is used to identify tissue characteristics, then time consumption is reduced and productivity increases, but the reliability may decrease without sufficient training data

Engineering Contradiction:

Improveprocessing efficiencyVSAvoidclassification accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system incorporates feedback mechanisms where automated classification results are reviewed and verified by pathologists, and these verified results are used to continuously retrain and improve the machine learning models. This closed-loop feedback ensures that productivity gains from automation do not compromise reliability, as the models learn from real-world expert validation and progressively improve their accuracy over time.

Inventive Principle:

Principle #23Feedback

Data Source

PatentEP3682419B1Few-shot learning based image recognition of whole slide image at tissue level

Publication Date: 2025.05.21 NANTOMICS LLC

Figure 1
Figure 2
Figure 3

AI summary

A computer implemented method of generating at least one shape of a region of interest in a digital image is provided. The method includes obtaining, by an image processing engine, access to a digital tissue image of a biological sample; tiling, by the image processing engine, the digital tissue image into a collection of image patches; obtaining, by the image processing engine, a plurality of features from each patch in the collection of image patches, the plurality of features defining a patch feature vector in a multidimensional feature space including the plurality of features as dimensions; determining, by the image processing engine, a user selection of a user selected subset of patches in the collection of image patches; classifying, by applying a trained classifier to patch vectors of other patches in the collection of patches, the other patches as belonging or not belonging to a same class of interest as the user selected subset of patches; and identifying one or more regions of interest based at least in part on the results of the classifying.