Lesion False Positive Removal Using Dual Operating Points

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current medical imaging analysis for liver lesions relies heavily on human expertise, leading to inefficiencies and errors due to the manual evaluation of medical images, with a need for improved automated image analysis mechanisms to accurately detect and classify lesions.

Innovation Solution

An AI pipeline employing multiple machine learning and deep learning models with dual operating points for false positive removal, optimizing both group and element levels to enhance the accuracy of lesion detection and classification in medical images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If automated machine learning models are used for lesion detection, then productivity is improved, but false positive rate increases

Engineering Contradiction:
Improveautomated lesion detection efficiencyVSAvoidfalse positive rate
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The system dynamically adjusts the operating point of the machine learning model based on the detected object's characteristics. For potential lesions, the model operates at a first operating point optimized for sensitivity, while for normal tissue, it operates at a second operating point optimized for specificity. This dynamic adjustment allows the system to maintain high productivity while reducing false positives by adapting the classification threshold in real-time.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The invention changes the operational parameters of the machine learning model by implementing multiple operating points with different sensitivity-specificity trade-offs. The system configures the model with at least a first operating point for detecting potential lesions and a second operating point for classifying normal tissue, thereby changing the parameter settings based on the context to improve both productivity and reliability.

Inventive Principle:
Principle #35Parameter changes

2Device complexity

If a single operating point is used for machine learning classification, then device complexity is reduced, but measurement precision deteriorates

Engineering Contradiction:
Improvemodel configuration simplicityVSAvoidlesion detection accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

Rather than using a static single operating point, the system implements dynamic parameter selection where the operating point is adjusted based on the input data characteristics. The processor determines whether a detected object is a potential lesion or normal tissue and applies the appropriate operating point, thereby improving measurement precision without requiring complex multi-model architectures.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The classification process is segmented into different stages with different operating points. The first operating point is used for initial lesion detection, and the second operating point is used for confirming normal tissue. This segmentation allows each stage to be optimized independently, improving overall precision while keeping the overall system structure relatively simple.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4248403B1Multiple operating point false positive removal for lesion identification
Publication Date: 2026.02.11 GUERBET SA
  • EP4248403B1 patent drawingFigure 1
  • EP4248403B1 patent drawingFigure 2
  • EP4248403B1 patent drawingFigure 3A~3C

AI summary

A false positive removal engine is provided. The false positive removal engine receives detected objects in one or more images. A machine learning classifier computer model, configured with first operational parameters to implement a first operating point, processes the received input to classify each detected object as being a true positive or a false positive to generate a first set of object classifications. If the first set is empty, the false positive removal engine outputs the first set as a filtered list of objects; otherwise the ML classifier computer model is configured with second operational parameters to implement a second operating point, different from the first operating point, which then processes the received input to classify each detected object and generate a second set of objects classified as true positive, which is output by the false positive removal engine as the filtered list of objects.