Transport Interface Cavity State Detection Under Variable Lighting

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods struggle to accurately classify and detect the status of cavities in transport interfaces for sample tubes under varying illumination and lighting conditions and distances, requiring time-consuming calibration.

Innovation Solution

A method using a convolutional neural network (CNN) model, trained on images of transport interfaces with varying lighting and distance, to categorize cavity states, allowing for efficient detection and classification of empty, holder-present, closed, and open tube conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional image processing methods are used to detect cavity states, then the system requires time-consuming calibration under different lighting conditions, but the detection accuracy is compromised when illumination varies

Engineering Contradiction:
Improvecavity state detection accuracyVSAvoidcalibration time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The convolutional neural network model is pre-trained on a diverse dataset of transport interface images captured under various lighting conditions, distances, and cavity states during the offline training phase. This preliminary action enables the model to learn robust features that are invariant to illumination changes, eliminating the need for time-consuming on-site calibration when the system is deployed in different environmental conditions.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The training dataset encompasses images with varying parameters including illumination intensity, lighting direction, distance from the transport interface, and different cavity states. By training the model across this parameter space, the system learns to generalize and maintain detection accuracy without requiring recalibration when these parameters change during operation.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If manual calibration is performed for each lighting condition and distance, then detection accuracy under varying conditions improves, but the complexity of the system increases

Engineering Contradiction:
Improvecavity state detection accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces manual calibration procedures and complex adaptive algorithms with a convolutional neural network model that has been pre-trained to handle varying conditions. This substitution of mechanical/manual processes with an intelligent model simplifies the system architecture while maintaining or improving detection accuracy across different lighting and distance conditions.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If traditional detection methods are used, then the initial setup is simpler, but adaptation to new sample tube types requires extensive re-calibration

Engineering Contradiction:
Improveadaptability to new sample tube typesVSAvoidre-training time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The training dataset includes images of various sample tube types with different characteristics. When new tube types are introduced, the model can be efficiently re-trained using a smaller dataset of images containing the new tube types, leveraging transfer learning to maintain performance while adapting to new configurations with minimal time investment.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3979131B1A method for determining at least one state of at least one cavity of a transport interface configured for transporting sample tubes
Publication Date: 2026.03.04 ROCHE DIAGNOSTICS GMBH
  • EP3979131B1 patent drawingFigure 1
  • EP3979131B1 patent drawingFigure 2~3
  • EP3979131B1 patent drawingFigure 4A~4B

AI summary

A method for determining at least one state of at least one cavity (114) of a transport interface (116) configured for transporting sample tubes (118) is disclosed. The method comprises: i) (166) capturing at least one image of at least a part of the transport interface (116) by using at least one camera (136); ii) (168) categorizing the state of the cavity (114) into at least one category by applying at least one trained model on the image by using at least one processing unit (140), wherein the trained model is being trained on image data of the transport interface (116), wherein the image data comprises a plurality of images of the transport interface (116) with cavities (114) in different states; iii) (170) providing the determined category of at least one cavity (114) via at least one communication interface (164).