Cell Image Analysis Apparatus for Automated Teaching Data Generation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In regenerative medicine, manually labeling cell images for machine learning to identify deviated cells or impurities is time-consuming and inefficient, making it challenging for industrial-scale production of pluripotent stem cells.

Innovation Solution

A cell image analysis apparatus that automatically generates teaching data by specifying removal target regions in cell images using image processing techniques, such as subtraction images, and generates training data sets for machine learning, reducing the need for manual labeling and enabling accurate identification of deviated cells.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual labeling of cell images is performed to generate teaching data for machine learning, then accurate identification of deviated cells can be achieved, but time consumption and labor effort increase significantly

Engineering Contradiction:

Improveidentification accuracyVSAvoidlabeling time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary image processing to generate subtraction images that highlight deviated cells before the machine learning model processes them. This preprocessing step automatically creates enhanced contrast images where deviated cells stand out, reducing the need for manual labeling while maintaining identification accuracy

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary processing step that generates subtraction images as a bridge between raw cell images and machine learning input. This intermediary representation automatically emphasizes deviated cells through image subtraction, serving as a mediator that reduces manual labeling requirements while preserving identification accuracy

Inventive Principle:

Principle #24Intermediary (Mediator)

2Quantity of substance

If manual labeling of cell images is performed to generate teaching data, then training data can be obtained, but productivity decreases due to time-consuming processes

Engineering Contradiction:

Improvetraining data volumeVSAvoiddata generation efficiency

Core Design Contradiction:

Quantity of substanceVSProductivity

Solution Approach 1:

The system enables self-service by allowing the machine learning model to automatically generate its own teaching data through the subtraction image process. The model processes raw images, identifies deviated cells, and generates labeled training data autonomously without requiring manual annotation, thus maintaining data volume while dramatically improving productivity

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent changes the parameter representation by transforming raw cell images into subtraction images through image processing. This parameter transformation automatically highlights deviated cells and generates suitable training data format, increasing productivity while maintaining the necessary training data volume for model development

Inventive Principle:

Principle #35Parameter changes

3Illumination intensity

If phase contrast microscopy is used to observe transparent cells, then cell morphology can be visualized, but determination accuracy varies due to observer skill dependency

Engineering Contradiction:

Improvecell visibilityVSAvoiddetermination consistency

Core Design Contradiction:

Illumination intensityVSMeasurement precision

Solution Approach 1:

The patent replaces the mechanical/subjective human observation process with an automated image processing system. Instead of relying on observer skill to interpret phase contrast images, the system uses subtraction image processing and machine learning to objectively identify deviated cells, maintaining cell visibility while eliminating determination variability

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system introduces subtraction images as an intermediary representation that transforms subjective visual determination into objective data processing. The subtraction process creates enhanced contrast images that serve as a mediator between the transparent cell morphology and automated identification, improving determination consistency while preserving cell visibility

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS11686721B2Cell image analysis apparatus, cell image analysis system, method of generating training data, method of generating trained model, training data generation program, and method of producing training data

Publication Date: 2023.06.27 SHIMADZU CORP

AI summary

A cell image analysis apparatus that can achieve less time and effort for labeling for generation of teaching data than in a conventional example is provided. The cell image analysis apparatus includes an image obtaining unit that obtains a cell image including a removal target that is obtained by a microscope for observation of a cell, a teaching data generator that specifies a removal target region including the removal target within the cell image by performing predetermined image processing and generates as teaching data for machine learning, a label image that represents a location of the removal target region in the cell image, and a training data set generator that generates a set of the cell image and the label image as a training data set to be used in machine learning.