AI Cell Analysis Method Resolving Ambiguous Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI algorithms struggle with accurately classifying cells when the probability of belonging to multiple types is similar, leading to decreased classification accuracy.
Innovation Solution
The method involves obtaining optical signals from a specimen using a measurement unit and analyzing set data using an artificial intelligence algorithm that calculates the relevance degree between different pieces of data to determine the type of analyte, thereby improving classification accuracy by distinguishing between ambiguous cell types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If an AI algorithm individually analyzes features of each cell to determine cell type, then the analysis can handle complex cell characteristics, but the classification accuracy decreases when multiple types have similar probabilities
Solution Approach 1:
The patent segments the cell analysis process into two distinct stages: first extracting individual cell features using AI algorithms, then performing pairwise comparisons between cells to resolve ambiguous classifications. This segmentation allows the system to maintain the ability to analyze complex characteristics while improving classification accuracy through the additional comparison stage.
Solution Approach 2:
The patent implements a feedback mechanism where the classification results from individual cell analysis are used to generate reference information, which then feeds back into the pairwise comparison process. This feedback loop enables the system to refine and improve classification accuracy by using previously determined classifications to resolve ambiguities in subsequent comparisons.
2Loss of information
If the AI algorithm outputs probability values for multiple cell types, then the analysis provides comprehensive type information, but it creates ambiguity when probabilities are similar and reduces identification reliability
Solution Approach 1:
The patent extracts the probability values and type information from individual cell analyses and uses them as reference information for pairwise comparisons. By taking out this information and using it in a systematic comparison process, the system maintains comprehensive type information while resolving ambiguities through structured comparison of multiple cells.
Solution Approach 2:
The patent adds a new dimension to the analysis by introducing pairwise comparisons between cells as a third level of processing, beyond individual feature analysis and probability calculation. This dimensional extension allows the system to resolve ambiguities in cell type identification by considering relationships between multiple cells simultaneously.
Data Source
AI summary
Disclosed is an analysis method for analyzing an analyte in a specimen, the analysis method including: obtaining first data corresponding to an optical signal obtained from the analyte; inputting set data composed of a plurality of pieces of the first data, to an artificial intelligence algorithm capable of calculating a relevance degree between the pieces of the first data; and determining a type of the analyte by using the relevance degree.