Lung CT CADx Scoring for Pre-Biopsy Disease Characterization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current CADx systems are limited in clinical utility as they require prior knowledge of malignancy and histological subtype, failing to provide meaningful outputs before biopsy, and are invasive, costly, and prone to misdiagnosis due to tumour heterogeneity and incomplete tissue sampling.

Innovation Solution

A CADx system using machine learning models, such as neural networks, processes CT images with clinical parameters to predict disease characterization and malignancy before biopsy, providing scores for histological subtype, genetic mutations, and PD-L1 expression, enhancing decision-making with non-invasive, comprehensive image analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If tissue biopsy is performed for disease characterisation, then diagnostic accuracy is improved, but patient invasiveness and procedural risk increase

Engineering Contradiction:
Improvediagnostic accuracyVSAvoidpatient invasiveness
Core Design Contradiction:
Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent replaces the mechanical tissue biopsy procedure with an image-based analysis system using machine learning models. The system processes medical images (CT, MRI, PET, etc.) to predict histological subtype, genetic mutations, and PD-L1 expression, substituting the need for physical tissue sampling while maintaining diagnostic utility.

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

Solution Approach 2:

The patent introduces medical images as an intermediary between the patient and the diagnostic process. Instead of directly extracting tissue, the system uses imaging data as a mediator to infer disease characteristics through machine learning, reducing direct patient intervention while preserving diagnostic information.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Loss of information

If tissue biopsy is performed for complete disease characterisation, then comprehensive diagnostic information is obtained, but sampling error due to tumour heterogeneity increases

Engineering Contradiction:
Improvecomprehensive diagnostic informationVSAvoidsampling accuracy
Core Design Contradiction:
Loss of informationVSReliability

Solution Approach 1:

The patent replaces physical tissue sampling with image-based analysis, eliminating sampling error entirely. The machine learning models analyze comprehensive imaging data that captures the entire tumor and its microenvironment, providing complete disease characterisation without the limitations of partial tissue sampling.

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

Solution Approach 2:

The patent creates a multi-functional diagnostic system that simultaneously predicts multiple disease characteristics (histological subtype, genetic mutations, PD-L1 expression) from a single image analysis, providing comprehensive diagnostic information without requiring multiple separate procedures or tissue samples.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Loss of information

If multiple biopsies are performed to overcome tumour heterogeneity, then diagnostic completeness is improved, but procedural cost and patient burden increase

Engineering Contradiction:
Improvediagnostic completenessVSAvoidprocedural complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent implements a single image analysis system that performs multiple diagnostic functions simultaneously - predicting histological subtype, genetic mutations, and PD-L1 expression all in one process. This eliminates the need for multiple separate biopsy procedures while achieving complete diagnostic characterisation.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent merges multiple diagnostic objectives into a single unified analysis framework. The machine learning model integrates prediction of various disease characteristics from comprehensive imaging data, combining what would otherwise require multiple separate procedures into one non-invasive assessment.

Inventive Principle:
Principle #5Merging (Combining)

4Object-affected harmful factors

If image-based disease characterisation is implemented, then non-invasive comprehensive analysis is achieved, but system complexity increases

Engineering Contradiction:
Improvepatient invasivenessVSAvoidsystem complexity
Core Design Contradiction:
Object-affected harmful factorsVSDevice complexity

Solution Approach 1:

The patent segments the complex diagnostic task into distinct machine learning models or prediction modules, each targeting specific disease characteristics (histological subtype prediction, genetic mutation detection, PD-L1 expression analysis). This modular architecture manages system complexity while providing comprehensive non-invasive characterisation.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4071768B1CAD device and method for analysing medical images
Publication Date: 2026.03.25 OPTELLUM LTD
  • EP4071768B1 patent drawingFigure 1
  • EP4071768B1 patent drawingFigure 2
  • EP4071768B1 patent drawingFigure 3

AI summary

A method for analysing images in a computer aided diagnosis system (CADx) to provide a first image analysis score and a second image analysis score for an image is described. The method comprising; receiving an input comprising at least one input image showing all or part of the lungs of a subject; analysing the input to calculating a first image analysis value and a second image analysis value for the input and processing the calculated values to generate corresponding first image analysis and second image analysis scores and outputting at least one of the first image analysis score and the second image analysis score for the subject. A computer aided diagnosis system (CADx) and a method of training a computer aided diagnosis system are also described.