Digital PCR Leakage Detection Using Partition Neighborhood Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Digital PCR systems face challenges in accurately identifying leakage between partitions due to overfilling, underfilling, or defects, leading to inaccurate fluorescence measurements and potential overestimation of target presence, which current manual methods are slow, laborious, and subjective.

Innovation Solution

A machine learning model trained on extensive input data sets is used to identify partitions with leakage by analyzing features of the neighborhood around each partition, including fluorescence intensity and analyte status, utilizing hexagonal partitions in a digital PCR system.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If manual inspection methods are used to detect leakage, then the detection process is simple to implement, but the detection speed is slow and labor-intensive

Engineering Contradiction:
Improvedetection speedVSAvoiddetection system complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent replaces manual visual inspection (mechanical human operation) with an automated machine learning-based image analysis system. The system uses trained models to automatically detect leakage patterns in dPCR plates, eliminating the need for manual inspection while significantly improving detection speed and consistency.

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

Solution Approach 2:

The system enables self-service detection by automatically analyzing images of dPCR plates and identifying leakage without requiring operator intervention. The machine learning model processes images autonomously, making the detection system self-sufficient and reducing labor requirements.

Inventive Principle:
Principle #25Self-service

2Measurement precision

If manual leakage detection is used, then the system is easy to operate, but the detection accuracy is low and subjective

Engineering Contradiction:
Improveleakage detection accuracyVSAvoiddetection method complexity
Core Design Contradiction:
Measurement precisionVSEase of operation

Solution Approach 1:

The patent replaces subjective manual judgment with objective machine learning-based analysis. The trained models consistently identify leakage patterns based on learned features from training data, eliminating operator subjectivity and improving detection accuracy and reproducibility.

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

Solution Approach 2:

The system uses feedback from training data to continuously improve detection accuracy. The machine learning models are trained on labeled examples of leakage and non-leakage patterns, allowing the system to learn from past detections and improve its accuracy over time through iterative optimization.

Inventive Principle:
Principle #23Feedback

3Reliability

If leakage is not detected, then the assay can proceed without interruption, but the diagnostic results become inaccurate due to overestimation of target presence

Engineering Contradiction:
Improvediagnostic accuracyVSAvoidassay throughput
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent performs preliminary detection of leakage before final diagnostic analysis. By identifying and flagging partitions with leakage early in the process, the system prevents contaminated data from affecting diagnostic results, ensuring reliability while maintaining workflow efficiency through automated detection.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4325444B1Digital PCR leakage detection and correction methods and systems
Publication Date: 2026.04.22 ROCHE DIAGNOSTICS GMBH
  • EP4325444B1 patent drawingFigure 1
  • EP4325444B1 patent drawingFigure 2
  • EP4325444B1 patent drawingFigure 3

AI summary

System and methods in this disclosure identify leakage from one partition to surrounding partitions. Detecting leakage may include input data describing the location of the partition, the fluorescence intensity, and whether the partition is positive for an analyte. A machine learning model may be trained on an extensive input data set to identify partitions that have leakage or a high probability of leakage. In some embodiments, features used in the machine learning model include values that describe a neighborhood around the partition and not merely the partition itself.