Blood Sample Classification Models Using Hybrid Concentration Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current optics-based methods for analyzing blood samples, such as determining optical density and microscopic measurements, face challenges in accurately classifying entities like platelets and white blood cells at varying concentrations, requiring different classifier performance levels for specificity and sensitivity.
Innovation Solution
A method using a computer processor to analyze microscopic images of blood samples with multiple classification models, switching between models based on concentration thresholds, employing a hybrid model for accurate identification and estimation of entities like platelets, and adjusting models dynamically during analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single classification model is used for analyzing blood samples, then the device complexity is low, but the classification accuracy varies significantly at different entity concentrations
Solution Approach 1:
The patent divides the classification task into multiple specialized models, each optimized for specific concentration ranges of blood entities. Instead of using one general model, the system segments the problem by creating distinct classifiers for different entity types (platelets, white blood cells, red blood cells) and concentration levels, thereby improving accuracy without requiring an excessively complex single model
Solution Approach 2:
The patent implements a dynamic classification system that adapts the chosen model based on real-time sample characteristics. The system dynamically selects which classification model to apply by analyzing sample properties first, then routing to the appropriate specialized model, making the overall system adaptable rather than static
2Measurement precision
If different classification models are used for different entity concentrations, then the classification accuracy improves, but the ease of operation decreases
Solution Approach 1:
The classification system performs self-service by automatically selecting the appropriate model based on sample characteristics without requiring manual intervention. The system autonomously analyzes the sample, determines the appropriate classification approach, and executes the correct model, thereby maintaining ease of operation while improving accuracy
3Reliability
If multiple classification models are implemented, then the reliability of entity identification improves, but the analysis time increases
Solution Approach 1:
The patent performs preliminary actions by first analyzing sample characteristics before applying the main classification task. This preliminary step allows the system to pre-determine which model will be most effective, avoiding the need to run multiple full classification processes and thereby reducing overall analysis time while maintaining high reliability
Data Source
AI summary
Apparatus and methods are described including analyzing one or more microscopic images of the blood sample using a machine-learning classifier. An entity within the one or more microscopic images is identified using a first classification model, and a first estimated concentration of the entity within the sample is determined, based upon the entity as identified using the first classification model. The entity is identified within the one or more microscopic images using a second classification model, and a second estimated concentration of the entity within the sample is determined, based upon the entity as identified using the second classification model. The first and second estimated concentrations are compared to each other, and, in response to the comparison, a hybrid classification model that is a hybrid of the first and second classification models is used. Other applications are also described.