Machine-Learned Cell Counting Algorithm

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods for estimating the number and degree of confluence of cells in cell cultures are inaccurate, time-intensive, and prone to errors, requiring manual counting or threshold-based techniques that require expert knowledge and slow down the analysis process.

Innovation Solution

A computer-implemented method that acquires light-microscope images, adapts the image size to a predefined reference value, and uses machine-learned algorithms to determine the number and degree of confluence of cells, enabling fully automated or partly automated evaluation with high accuracy and computational efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual counting techniques are used to determine the number of cells, then the accuracy of cell counting is improved, but the time consumption and complexity of the analysis process increase

Engineering Contradiction:

Improveaccuracy of cell countingVSAvoidtime consumption

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces manual mechanical counting with an automated image analysis system using machine-learned algorithms. The processor automatically analyzes microscopy images to count cells and determine confluence, eliminating the need for manual intervention while maintaining high accuracy through trained neural networks that have learned cell patterns from training data.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs self-service by automatically analyzing images without requiring expert knowledge for parameter setting. The machine-learned algorithms autonomously process the images, count cells, and determine confluence degrees, making the analysis process independent of operator expertise and significantly reducing time consumption.

Inventive Principle:

Principle #25Self-service

2Loss of time

If threshold value-based automated techniques are used for cell analysis, then the time consumption is reduced, but the reliability and accuracy of the estimation deteriorate due to parameter setting requirements

Engineering Contradiction:

Improvetime consumptionVSAvoidreliability of estimation

Core Design Contradiction:

Loss of timeVSReliability

Solution Approach 1:

The patent fundamentally changes the parameters from fixed threshold values to learned parameters from training data. Instead of using predetermined contrast thresholds, the system uses machine-learned algorithms that have adapted to specific cell types and imaging conditions, making the analysis both rapid and reliable without requiring expert parameter setting.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces the mechanical threshold-based classification system with an intelligent machine-learning system. The neural networks automatically learn optimal decision boundaries from training data, eliminating the need for manual threshold setting while improving reliability through adaptive learning from actual cell images.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Extent of automation

If threshold value-based techniques are used for cell analysis, then automation is achieved, but the device complexity increases due to parameter setting requirements

Engineering Contradiction:

Improveautomation of analysisVSAvoidcomplexity of algorithm parameterization

Core Design Contradiction:

Extent of automationVSDevice complexity

Solution Approach 1:

The system achieves self-service automation where the machine-learned algorithms automatically handle all aspects of image analysis without requiring user intervention for parameter setting. The algorithms independently process images, count cells, and determine confluence, simplifying the user interface while maintaining high automation through pre-trained models.

Inventive Principle:

Principle #25Self-service

4Measurement precision

If manual estimation techniques are used for confluence determination, then the accuracy is improved, but the productivity of the analysis process decreases

Engineering Contradiction:

Improveaccuracy of confluence estimationVSAvoidproductivity of analysis

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent replaces manual visual estimation with automated image processing using machine-learned algorithms. The system rapidly analyzes microscopy images to determine confluence degrees by comparing pixel intensities and patterns against learned models, achieving both high accuracy and rapid processing that manual methods cannot provide.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables continuous automated analysis of multiple images without interruption, maintaining high productivity while preserving accuracy through consistent application of machine-learned algorithms. The processor can continuously process images in sequence, providing rapid results for multiple samples without the intermittent nature of manual estimation.

Inventive Principle:

Principle #20Continuity of useful action

Data Source

PatentUS20230105948A1Training a machine-learned algorithm for cell counting or for cell confluence determination

Publication Date: 2023.04.06 CARL ZEISS MICROSCOPY GMBH

AI summary

Various examples of the disclosure relate to aspects associated with training a machine-learned algorithm configured to count cells in a microscopy image or to determine a degree of confluence of the cells.