Sample Container Identification Network Retraining With Core Data Sets

Solution Overview

Problem

Existing diagnostic laboratory systems face high training costs and inefficiencies when adding new types of sample containers due to the need for large amounts of training data to retrain machine learning models, and the systems struggle to identify sample containers under real-world conditions that differ from controlled imaging settings.

Innovation Solution

A method of training and retraining sample container identification networks using smaller core data sets that reflect current conditions, allowing for efficient adaptation to new container types and real-world variations, and a system comprising an imaging device, memory, and computer to capture and process images for network training and retraining.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If large amounts of training data are used to retrain machine learning models for new sample container types, then identification accuracy is improved, but training cost and time increase excessively

Engineering Contradiction:

Improvesample container identification accuracyVSAvoidtraining time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary actions by continuously capturing and storing images of sample containers in real-world conditions during normal operation. This pre-collected visual data is stored in a database and can be used for rapid model retraining when new container types are introduced, eliminating the need to collect large datasets from scratch and significantly reducing training time while maintaining identification accuracy

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If large amounts of training data are used to retrain machine learning models for new sample container types, then identification accuracy is improved, but training cost increases excessively

Engineering Contradiction:

Improvesample container identification accuracyVSAvoidtraining cost

Core Design Contradiction:

Measurement precisionVSLoss of energy

Solution Approach 1:

The system performs preliminary actions by continuously capturing and storing images of sample containers in real-world conditions during normal operation. This pre-collected visual data is stored in a database and can be used for rapid model retraining when new container types are introduced, eliminating the need to collect large datasets from scratch and significantly reducing training time while maintaining identification accuracy

Inventive Principle:

Principle #10Preliminary action

3Reliability

If machine learning models are trained on controlled imaging settings, then initial identification performance is good, but the system struggles to identify sample containers under real-world conditions

Engineering Contradiction:

Improveidentification performance under real-world conditionsVSAvoidadaptability to different imaging conditions

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The system performs preliminary actions by continuously capturing and storing images of sample containers in real-world conditions during normal operation. This pre-collected visual data is stored in a database and can be used for rapid model retraining when new container types are introduced, eliminating the need to collect large datasets from scratch and significantly reducing training time while maintaining identification accuracy

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements feedback by continuously monitoring identification results and using failed identification cases to update and retrain the machine learning model. When the model fails to identify a sample container, the image is retrieved from the database and used to retrain the model, creating a closed-loop system that continuously improves adaptability to real-world conditions based on actual performance feedback

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20260002952A1Devices and methods for training sample container identification networks in diagnostic laboratory systems

Publication Date: 2026.01.01 SIEMENS HEALTHCARE DIAGNOSTICS INC

AI summary

A method of training a sample container identification network of a diagnostic laboratory system includes obtaining a plurality of data subsets, wherein each data subset is smaller than a full training data set used to train the sample container identification network and includes a plurality of images of one or more sample containers. The sample container identification network is trained on each of the plurality of data subsets to generate a plurality of trained sample container identification networks. Each of the trained sample container identification networks are testing using testing data that includes test images of sample containers, wherein the testing includes identifying the sample containers in the test images. A core data set is selected from one of the plurality of data subsets based on the testing. the core data set for use in training a deployed sample container identification network. Other methods and systems are disclosed.