Automated Brain Segmentation via Inflection Point Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for segmenting human brain tissues from MRI images require manual adjustment of parameters, leading to inefficiencies and user bias, especially in large datasets, due to varying MRI image qualities and subject-specific variations.

Innovation Solution

A computer-implemented method using the BET brain extraction tool (BET) with a fractional constant b t to compute multiple surface models, fitting a curve to determine an inflection point, and identifying a corresponding b̃ t for automated segmentation, reducing the need for manual tuning.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual parameter adjustment is used to optimize segmentation quality, then segmentation accuracy can be improved, but time consumption and user bias increase significantly

Engineering Contradiction:
Improvesegmentation accuracyVSAvoidtime consumption
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs self-optimization by automatically analyzing the relationship between fractional constant parameters and brain tissue volume, identifying optimal parameters without human intervention. The algorithm processes multiple test runs, fits curves to the data, and determines inflection points autonomously, eliminating the need for manual parameter tuning while maintaining high segmentation accuracy

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system conducts preliminary automated testing and analysis before final segmentation. Multiple test runs with different fractional constant values are performed in advance, curve fitting is conducted, and optimal parameters are predetermined through inflection point detection, so that the actual segmentation can proceed efficiently without manual intervention

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If manual parameter adjustment is performed to achieve best fit segmentation, then segmentation quality improves, but user bias is introduced

Engineering Contradiction:
Improvesegmentation qualityVSAvoiduser bias
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The segmentation system serves itself by automatically determining optimal parameters through objective mathematical analysis. The algorithm independently processes test results, fits curves to volume-data, and identifies inflection points without human input, completely eliminating user bias while maintaining high segmentation quality through reproducible, objective criteria

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system implements automated feedback loops where segmentation results are continuously evaluated against volume measurements. The algorithm uses this feedback to refine parameter selection through curve fitting and inflection point detection, replacing subjective human judgment with objective, bias-free automated decision-making

Inventive Principle:
Principle #23Feedback

3Measurement precision

If multiple test runs are conducted to find optimal parameters, then segmentation accuracy improves, but productivity decreases

Engineering Contradiction:
Improvesegmentation accuracyVSAvoidprocessing efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The system replaces the mechanical process of manual parameter tuning with automated computational analysis. Instead of users sequentially adjusting parameters based on visual inspection, the algorithm automatically conducts multiple test runs, numerically fits curves to the data, and computationally determines inflection points, dramatically improving processing efficiency while maintaining accuracy

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

Solution Approach 2:

The system autonomously manages the entire multi-run optimization process without human intervention. It automatically executes multiple test runs with different parameters, collects volume data, performs curve fitting, and identifies optimal parameters through inflection point detection, transforming a previously time-consuming manual process into an efficient automated workflow

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4141791B1Automatic brain model extraction
Publication Date: 2026.04.01 DASSAULT SYSTEMES SA
  • EP4141791B1 patent drawingFigure 1
  • EP4141791B1 patent drawingFigure 2~3
  • EP4141791B1 patent drawingFigure 4~5

AI summary

The disclosure notably relates to a method of segmentation of a medical image of a human brain. The method comprises providing a voxelized 3D medical image of the human brain, computing at least two surface models of the human brain using a BET method, each surface model being computed for a unique fractional constant bt. For each computed surface model, determining a volume (Volume_bt) comprised in the computed surface model, thereby obtaining a set of sample pairs (bt, Volume_ bt), fitting a curve to the sample pairs (bt, Volume_bt) of the set, determining an inflection point of the curve, identifying a fractional constant bt̃ corresponding to the determined inflection point, and computing the surface model of the human brain using the BET method that is dependent of bt̃.