Deep Learning Cell Morphology Analysis for Rare Blood Cell Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for cell identification, including microscopic observation and machine learning, face challenges in accurately identifying cells with low emergence frequencies and require significant labor for training data generation, limiting analysis accuracy and generalization capability.

Innovation Solution

An image analysis method using a deep learning algorithm with a neural network structure to calculate the probability of a cell belonging to specific morphology classifications, enabling identification of cell types and abnormalities without manual observation, utilizing training data generated from stained images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If machine learning techniques are used for cell identification, then automation is improved, but training data generation requires tremendous labor

Engineering Contradiction:

Improveautomation of cell identificationVSAvoidtime for training data generation

Core Design Contradiction:

Extent of automationVSLoss of time

Solution Approach 1:

The system performs preliminary action by automatically generating training data through simulated microscope imaging of digitized pathology slides. The training data is prepared in advance with automated cell segmentation, feature extraction, and label assignment, eliminating the need for manual annotation of training datasets.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system creates copies by generating synthetic training data that replicates real microscopic images. Virtual slides are created from digital pathology images, and simulated microscope images are generated with added noise and artifacts to match real-world conditions, providing abundant training data without manual collection.

Inventive Principle:

Principle #26Copying

2Measurement precision

If the number of training data is increased to improve analysis accuracy, then generalization capability is improved, but the labor required for data creation increases

Engineering Contradiction:

Improveanalysis accuracyVSAvoidease of training data creation

Core Design Contradiction:

Measurement precisionVSEase of manufacture

Solution Approach 1:

The system generates large volumes of training data by copying and transforming digital pathology images into simulated microscope images. Multiple variations are created through adding different types of noise, adjusting contrast, and simulating various imaging conditions, efficiently producing diverse training datasets without manual effort.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The system applies parameter changes by modifying image characteristics such as noise levels, contrast, brightness, and resolution to generate diverse training samples from a single source. This creates data variation that improves model generalization while maintaining automated generation efficiency.

Inventive Principle:

Principle #35Parameter changes

3Productivity

If flow-type automatic hemocyte classification apparatus is used to increase examination capacity, then productivity is improved, but identification accuracy for low emergence frequency cells deteriorates

Engineering Contradiction:

Improveexamination capacityVSAvoididentification accuracy for rare cells

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system segments the analysis process into distinct stages: digital slide scanning, automated cell segmentation, feature extraction, and deep learning-based classification. This segmentation allows specialized algorithms to handle each task optimally, improving accuracy for rare cell identification while maintaining high throughput capacity.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system replaces mechanical flow cytometry with digital image analysis and deep learning algorithms. This substitution enables detailed morphological analysis of individual cells at high magnification, improving detection of rare cell types while maintaining automated high-throughput processing capability.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS20250209621A1Image analysis method, apparatus, non-transitory computer readable medium, and deep learning algorithm generation method

Publication Date: 2025.06.26 SYSMEX CORP

AI summary

Disclosed is an image analysis method including inputting analysis data into a classifier having a neural network structure, the analysis data being generated from an image of an analysis target cell and including information regarding the analysis target cell; and classifying, by use of the classifier, the analysis target cell into at least one of categories of a blood cell. The categories include morphological features of white blood cells. The morphological features of white blood cells include at least morphological nucleus abnormality, presence of vacuole, granule morphological abnormality, granule distribution abnormality, presence of abnormal granule, cell size abnormality, presence of inclusion body, or bare nucleus.