Parametric Level Set Framework for Lesion Segmentation and Registration

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods fail to simultaneously segment and register lesions in medical images acquired at different phases and times, particularly in oncology follow-up studies, due to varying contrast and deformations caused by radiation treatment, breathing motion, and patient position changes.

Innovation Solution

A novel method and apparatus that use a processor to optimize segmentation and registration of objects in baseline and follow-up images by evolving a level set function to maximize a conditional probability expression, adjusting parametric level set functions, and generating segmented and registered images for display.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If current methods are used for segmentation and registration, then segmentation can be performed in baseline and follow-up volumes, but simultaneous segmentation and registration of lesions across different phases and times cannot be achieved

Engineering Contradiction:

Improveautomatic segmentation and registrationVSAvoidcomplexity of segmentation and registration framework

Core Design Contradiction:

Extent of automationVSDevice complexity

Solution Approach 1:

The patent combines segmentation and registration into a single unified framework that operates simultaneously across multiple phases and time points. Instead of performing segmentation first and then registration separately, the method integrates both operations into one cohesive process that handles baseline and follow-up volumes together, enabling automatic lesion correspondence establishment without requiring separate processing steps

Inventive Principle:

Principle #5Merging (Combining)

2Measurement precision

If lesions are segmented in all phases of follow-up studies, then comprehensive cancer monitoring is achieved, but the ability to detect and segment lesions becomes extremely challenging due to heavily varying contrast between phases

Engineering Contradiction:

Improvelesion detection accuracyVSAvoidlesion segmentation difficulty

Core Design Contradiction:

Measurement precisionVSDifficulty of detecting and measuring

Solution Approach 1:

The method performs preliminary segmentation in the baseline volume to establish reference lesion characteristics and positions. These baseline segmentations are then used to guide and constrain the segmentation process in follow-up phases, making the algorithm more robust to contrast variations. By having the baseline segmentation ready in advance, the system can automatically adapt to phase-specific challenges without requiring manual intervention in each phase

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The unified framework incorporates feedback mechanisms where segmentation results from one phase inform and improve segmentation in other phases. The registration component uses correspondence information from baseline to guide follow-up segmentation, and vice versa, creating a feedback loop that progressively refines lesion detection across all phases despite varying contrast conditions

Inventive Principle:

Principle #23Feedback

3Measurement precision

If manual segmentation and correspondence establishment are performed, then accurate lesion identification can be achieved, but the process is time-consuming and cannot be performed automatically

Engineering Contradiction:

Improvelesion correspondence accuracyVSAvoidtime for segmentation and registration

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs self-service by automatically establishing lesion correspondence between baseline and follow-up volumes without requiring manual intervention. The unified framework uses the parametric level set functions and optimization algorithms to autonomously identify and track lesions across phases, eliminating the need for operators to manually segment and match lesions while maintaining high accuracy through the mathematical constraints and prior knowledge embedded in the model

Inventive Principle:

Principle #25Self-service

4Adaptability or versatility

If follow-up studies include multiple contrast phases, then comprehensive cancer monitoring is enabled, but the number of volumes to process increases to 8 or more volumes

Engineering Contradiction:

Improvecancer monitoring capabilityVSAvoidprocessing efficiency

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The unified segmentation and registration framework is designed to be universal and handle multiple phases and time points simultaneously within a single processing operation. Rather than requiring separate segmentation and registration pipelines for each phase, the method uses a multi-functional approach that processes all 8 or more volumes in one unified operation, improving productivity while maintaining the ability to monitor cancer across all contrast phases

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS8064673B2Combined segmentation and registration framework for parametric shapes

Publication Date: 2011.11.22 SIEMENS HEALTHINEERS AG

AI summary

Progress monitoring of lesions is done automatically by segmentation and registration of lesions in multi-phase medical images. A parametric level-set framework includes a model optimization for any number of lesions. The user specifies lesions in a baseline volume by clicking inside of them. The apparatus segments the lesions automatically in the baseline and follow-up volumes. The segmentation optimization compensates for lesion motion between baseline and follow-up volumes. 2D and 3D medical patient data can be processed by the methods.