CT Lung Segmentation and Weighted Scoring for ILD Progression

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for measuring lung function and predicting the progression of interstitial lung diseases (ILDs) are plagued by variability and ethical challenges, making it difficult to design effective clinical trials and identify therapeutic responses.

Innovation Solution

A computer-implemented method that analyzes CT scans to quantify ILD extent and predict progression by segmenting lung images, applying weightings to identified structures based on their relative position to the lung periphery, and using a machine learning model to identify reticulo-vascular structures.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If forced vital capacity (FVC) testing is used to measure lung function and predict ILD progression, then lung function can be quantified, but the results suffer from high variability and cannot reliably identify patients whose disease is progressing with stable lung function

Engineering Contradiction:
ImproveILD progression prediction accuracyVSAvoidFVC test variability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent segments the lung parenchyma into multiple regions (e.g., upper, middle, lower zones and peripheral, central regions) and analyzes reticular pattern density in each segment separately. This segmentation allows the system to capture localized disease progression that may be missed by global FVC measurements, thereby improving prediction accuracy while reducing the impact of overall lung function variability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different analysis methods and weighting factors to different lung regions based on their specific characteristics. Reticular patterns are detected and quantified with region-specific thresholds and criteria, allowing the system to account for the fact that ILD affects different parts of the lung differently. This local quality approach enables more precise progression detection in affected areas while ignoring stable regions, resolving the contradiction between measurement precision and reliability.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If traditional radiologist assessment or lung function metrics are used to quantify ILD extent, then disease measurement is possible, but the methods lack standardization and cannot consistently identify patients suitable for clinical trials

Engineering Contradiction:
ImproveILD extent quantification accuracyVSAvoidAssessment method standardization
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent implements an automated computational system that performs reticular pattern detection and quantification without requiring manual radiologist interpretation. The algorithm automatically segments lung regions, detects reticular structures, applies weighting factors, and generates standardized scores. This self-service automation eliminates inter-observer variability and ensures consistent application of assessment criteria across all patients, thereby improving both measurement precision and standardization while reducing methodological complexity.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent transforms qualitative radiological assessments into quantitative parameters by measuring reticular pattern density, area, and distribution in standardized lung regions. The system converts visual patterns into numerical scores that can be objectively compared across patients and time points. This parameter transformation enables precise ILD extent quantification and provides standardized metrics for identifying clinical trial candidates, resolving the contradiction between measurement accuracy and assessment standardization.

Inventive Principle:
Principle #35Parameter changes

3Loss of information

If comprehensive lung imaging analysis is performed to identify all structures and patterns, then detailed disease characterization is achieved, but the analysis complexity and computational requirements increase significantly

Engineering Contradiction:
ImproveDisease pattern detection completenessVSAvoidImage analysis complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent extracts and focuses specifically on reticular patterns as the key indicator of ILD progression, rather than attempting to analyze all possible lung structures and patterns. The system uses image processing techniques to isolate and detect reticular structures specifically, applying targeted algorithms that identify these patterns while ignoring other lung features. This extraction approach maintains comprehensive disease pattern detection for the most relevant features while significantly reducing overall analysis complexity and computational requirements.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP4246434B1Computer implemented method for quantifying and predicting the progression of interstitial lung disease
Publication Date: 2025.07.02 BRAINOMIX LTD
  • EP4246434B1 patent drawingFigure 1~2
  • EP4246434B1 patent drawingFigure 3
  • EP4246434B1 patent drawingFigure 4A~4C

AI summary

A computer implemented method for quantifying and predicting the progression of interstitial lung disease is disclosed herein. The method comprises obtaining at least one image based on a scan of at least a part of a patient's lung, segmenting the image to obtain a lung mask defining the periphery of the lung and to identify structures within the lung mask, and applying a weighting to the identified structures based on the relative position of the structures relative to the lung periphery to obtain a weighted score. The method may then comprise quantifying the extent of interstitial lung disease, and/or predicting the progression of interstitial lung disease, based on the weighted score.