Ultrasound Liver Segmentation for Inhomogeneous Fat Quantification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current liver fat quantification algorithms provide inaccurate measurements for patients with inhomogeneous liver fat distribution, as they assume homogeneous fat distribution, leading to unreliable results and increased operator workload.

Innovation Solution

An ultrasound imaging system that segments liver regions, extracts parameters such as image homogeneity and speckle size, and uses a machine learning model to classify fat distribution as homogeneous or inhomogeneous, providing visual or auditory warnings for accurate ROI selection.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If current liver fat quantification algorithms assume homogeneous fat distribution, then measurement simplicity is maintained, but measurement precision deteriorates for patients with inhomogeneous liver fat

Engineering Contradiction:

Improvemeasurement simplicityVSAvoidliver fat measurement accuracy

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The liver region is segmented into multiple sub-regions or zones based on fat distribution patterns. The algorithm divides the liver into homogeneous and inhomogeneous areas, allowing separate analysis of each segment. This segmentation enables accurate detection of inhomogeneous fat distribution while maintaining overall measurement simplicity through automated region classification.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The algorithm changes the measurement parameters dynamically based on detected fat distribution patterns. When inhomogeneous distribution is detected, the system adjusts the quantification approach by using multiple local measurements or weighted averaging schemes instead of a single global measurement, thereby improving precision without significantly increasing operational complexity.

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If automated classification models are implemented to identify inhomogeneous liver fat, then measurement precision is improved, but device complexity increases

Engineering Contradiction:

Improveliver fat measurement accuracyVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs self-classification by automatically analyzing ultrasound images and identifying inhomogeneous fat patterns without requiring operator intervention or manual assessment. The automated classification model processes images independently, reducing the need for complex user interfaces or manual calibration procedures, thereby limiting the increase in operational complexity despite improved measurement precision.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

Manual visual inspection and operator judgment are replaced with an automated computational classification model. The system uses machine learning algorithms to automatically distinguish homogeneous from inhomogeneous fat distribution patterns, substituting human expertise with an automated digital system that improves precision while keeping the user interface simple.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If operators perform manual visual inspection to identify inhomogeneous liver fat, then adaptability to individual cases is improved, but productivity decreases due to increased workload

Engineering Contradiction:

Improvecase-by-case assessment capabilityVSAvoidmeasurement throughput

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The system automatically performs case-by-case assessment by analyzing each patient's ultrasound images through the classification model. The automated system adapts to individual cases by detecting unique fat distribution patterns without requiring operator time or expertise for manual evaluation, thereby maintaining high adaptability while significantly increasing measurement throughput and productivity.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

Manual operator inspection is replaced with an automated image analysis system that processes ultrasound images rapidly. The computational model provides case-specific assessments by identifying inhomogeneous patterns in each patient's data, substituting slow manual evaluation with fast automated processing that maintains diagnostic accuracy while improving workflow efficiency.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP4320587B1Systems, methods, and apparatuses for identifying inhomogeneous liver fat

Publication Date: 2025.12.24 KONINKLIJKE PHILIPS NV

Figure 1
Figure 2
Figure 3

AI summary

An ultrasound imaging system may acquire an image of a liver. The liver may be segmented from the image. Parameters, such as image homogeneity map, intensity probability chart, and/or speckle size diagram, may be extracted from the liver portion of the image. The parameters may be used to determine whether fatty liver deposits are diffuse or inhomogeneous. In some examples, inhomogeneous regions may be excluded from the calculation of liver fat quantification measurements. In some examples, the inhomogeneous regions may be displayed so that a user may select a region of interest that excludes the inhomogeneous regions to calculate the liver fat quantification measurements.