Cell Proportion Estimation From Images Without Sample Damage

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods struggle to accurately determine the proportion of specific cell types or microorganisms in a sample without causing damage or altering their state through destructive processes or staining.

Innovation Solution

An apparatus and method that utilize image analysis and machine learning to estimate the proportion of detection targets in a sample based on a reference sample, using parameters such as shape and texture, without destructive or staining processes, and employing a learning model to enhance accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If destructive processes or staining are used to determine cell proportions, then measurement precision is improved, but the cells or microorganisms are damaged or altered

Engineering Contradiction:
Improveproportion determination accuracyVSAvoidcell damage or state alteration
Core Design Contradiction:
Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent replaces destructive mechanical/chemical methods (staining, destruction) with optical imaging and machine learning analysis. The image capturing unit captures optical images of cells, and the estimation unit uses learned models to determine proportions without physical or chemical intervention, thus maintaining cell integrity while achieving accurate measurement

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces an intermediary learning model that acts as a mediator between the image data and the proportion determination. Instead of directly using destructive methods, the system uses trained neural networks to interpret image features and infer cell proportions, enabling non-invasive measurement with high accuracy

Inventive Principle:
Principle #24Intermediary (Mediator)

2Object-affected harmful factors

If image analysis without destructive processes is used, then cell integrity is maintained, but measurement precision deteriorates

Engineering Contradiction:
Improvecell integrityVSAvoidproportion determination accuracy
Core Design Contradiction:
Object-affected harmful factorsVSMeasurement precision

Solution Approach 1:

The patent applies preliminary action by pre-training the learning model using supervised learning with labeled training data. The model is trained in advance on images with known cell proportions, enabling it to accurately estimate proportions in new samples without requiring destructive verification during actual measurement, thus maintaining both cell integrity and measurement precision

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent substitutes traditional image analysis methods with advanced machine learning models that can extract subtle features from optical images. The neural network learns complex patterns in cell morphology and arrangement, enabling precise proportion determination from non-invasive images that would be insufficient for traditional analysis methods

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP4614464A1Apparatus, method, and program
Publication Date: 2025.09.10 YOKOGAWA ELECTRIC CORP
  • EP4614464A1 patent drawingFigure 1
  • EP4614464A1 patent drawingFigure 2
  • EP4614464A1 patent drawingFigure 3

AI summary

Provided is an apparatus, including an acquisition unit that acquires an image of a sample containing cells or microorganisms as image-capturing targets; an estimation unit that estimates a proportion of detection targets to image-capturing targets in a sample shown in an image acquired by the acquisition unit, based on a group of an image of a reference sample containing image-capturing targets and a proportion of cells or microorganisms as the detection targets to the image-capturing targets in the reference sample; and an output unit that outputs a proportion estimated by the estimation unit; and a determination unit that determines an extraction condition extracted in an image of the group by image-capturing targets according to the proportion of the group, among extraction conditions of image-capturing targets, which are settable by a parameter indicating a shape or a texture of an image-capturing target in the image, based on the group of an image of the reference sample and the proportion of the detection targets to the image-capturing targets contained in the reference sample, wherein the estimation unit estimates the proportion regarding a sample shown by an input image, by using the extraction condition determined by the determination unit.