Image Classification Device for Morphological Feature Data Display
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for classifying differentiated and undifferentiated cell states based on morphological features lack a method for creating initial classification data and reclassifying results using user input, leading to reliance on specialist knowledge for medical applications.
Innovation Solution
An image classification device and method that includes an image input unit, image display unit, image analysis unit, feature amount calculation, extraction and sorting processing, and user classification input, allowing users to input classification destinations and display results, enabling efficient creation of initial classification data and comprehensive evaluation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automatic classification is performed using a classification model, then classification speed and productivity are improved, but reliability and accuracy deteriorate because the results cannot be fully trusted without specialist verification
Solution Approach 1:
The patent introduces an extraction and sorting unit as an intermediary between the automatic classification model and the final determination. This unit extracts specific feature amounts (such as circularity, area, diameter) and sorts cells based on these features, providing a structured intermediate representation that specialists can verify. The intermediary transforms the black-box classification output into interpretable feature-based groups that maintain both automation efficiency and specialist oversight capability.
Solution Approach 2:
The patent segments the classification process into distinct stages: automatic classification by model, extraction of morphological features, sorting based on features, and final specialist verification. This segmentation allows each stage to contribute its strength - the model provides speed, the feature extraction provides objectivity, and the specialist provides reliability - while maintaining overall system productivity and trustworthiness.
2Reliability
If specialists perform manual classification and evaluation, then reliability and accuracy are improved, but time consumption and loss of time increase
Solution Approach 1:
The patent implements preliminary action by having the extraction and sorting unit prepare classified groups and feature量 before specialist verification is needed. The system pre-extracts morphological features, pre-sorts cells based on these features, and pre-organizes data for verification, so that specialists receive ready-to-review structured information rather than raw unprocessed data, significantly reducing their time investment while maintaining accuracy.
Solution Approach 2:
The system performs self-service classification for routine tasks through the automatic classification model and feature-based sorting, handling the bulk of classification work autonomously. This self-service capability handles standard cases efficiently, freeing specialists to focus only on verification and edge cases, thereby reducing overall time consumption while preserving reliability through selective human involvement.
3Measurement precision
If comprehensive feature analysis is performed, then measurement precision is improved, but device complexity increases due to multiple processing units
Solution Approach 1:
The extraction and sorting unit serves multiple functions: it extracts morphological features, sorts cells based on these features, prepares data for verification, and provides objective criteria for classification. This multi-functionality consolidates what could be separate complex modules into a single versatile unit, achieving comprehensive feature analysis while limiting the growth of device complexity through functional integration.
Data Source
AI summary
Provided is an image classification device that facilitates efficient creation of teacher data and comprehensive evaluation on a basis of knowledge and experience of the user. The image classification device includes: an image input unit that acquires an image; an image display unit that displays the image acquired by the image input unit; an image analysis unit that calculates a feature amount from the acquired image; a feature amount display unit that displays the calculated feature amount; an extraction and sorting condition input unit that specifies an extraction and sorting condition with regard to the feature amount; an extraction and sorting processing unit that performs extraction and sorting processing based on the condition; an extraction and sorting processing result display unit that displays a result; an user classification input unit that allows a user to input a classification destination with regard to the image; and a user classification result display unit that displays an classification input content.