Image Analysis Device Center Determination for Cell Classification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In pathological diagnosis and automatic cell analysis using image analysis devices, cells located off-center in partial images often cannot be accurately classified, leading to inaccuracies in classification results.

Innovation Solution

The image analysis device employs a center determination process using learning data to identify the center of each partial image relative to the cell object, and if the center is off, it generates binary image data with marks to re-identify and classify the cell using corrected partial images, thereby improving classification accuracy and preventing multiple classifications of the same cell.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If partial image data is used for cell classification, then processing efficiency is improved, but classification accuracy deteriorates when cell is off-center

Engineering Contradiction:
Improveprocessing efficiencyVSAvoidclassification accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The system performs preliminary center determination before classification by analyzing the positional relationship between the cell object and partial image center. This preliminary action identifies whether the cell is off-center, allowing the system to then perform corrective actions (re-positioning or using alternative partial images) to ensure accurate classification while maintaining efficient processing.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system implements a feedback mechanism where the center determination result is used to control subsequent classification processing. When off-center cells are detected, the system provides feedback to adjust the processing approach (such as re-positioning the cell to center or selecting different partial images), thereby improving classification accuracy without significantly compromising processing efficiency.

Inventive Principle:
Principle #23Feedback

2Measurement precision

If center determination process is added, then classification accuracy is improved, but device complexity increases

Engineering Contradiction:
Improveclassification accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system uses the existing cell image data and object detection results to perform center determination without requiring additional specialized hardware or complex external systems. The determination unit leverages already-acquired image information and detection results to calculate positional relationships, making the center determination process self-contained and avoiding significant increases in device complexity.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The determination unit is designed to perform multiple functions: it not only determines whether the cell center matches the partial image center, but also identifies off-center cells and controls subsequent classification processing. This multi-functionality reduces the need for separate dedicated components, thereby limiting the increase in device complexity while achieving improved classification accuracy.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3624053B1Image analysis device
Publication Date: 2022.12.21 NAGASAKA TORU
  • EP3624053B1 patent drawingFigure 1
  • EP3624053B1 patent drawingFigure 2
  • EP3624053B1 patent drawingFigure 3A~3C

AI summary

An image analysis device may include a memory storing learning data for executing image analysis, and may obtain cell image data representing a cell image including a plurality of cell objects, sequentially identify plural pieces of partial image data from the cell image data, sequentially execute a center determination process on each of the plural pieces of partial image data, classify at least one cell corresponding to at least one cell object among a plurality of cell objects by using results of the center determination process on the plural pieces of partial image data and classification data included in learning data, and output a classification result.