Medical Image Analysis Using Quantitative Parameters for Reliable Case Retrieval

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current medical image analysis technologies face challenges in retrieving clinically significant similar cases due to the lack of reliable, clinically meaningful data and the risk of overfitting or acquiring insignificant patterns from limited data.

Innovation Solution

A medical image analysis method that uses quantitative parameters to segment anatomical regions, generate region-specific quantitative parameters, and store them in a database for effective similar case retrieval, allowing for the calculation of similarity and retrieval of clinically significant cases based on these parameters.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If artificial intelligence algorithms are applied to process medical images and data, then the capability to assist clinical decision making is improved, but the risk of overfitting and acquiring clinically insignificant patterns increases when data is limited

Engineering Contradiction:

Improveautomated medical image analysisVSAvoidclinical significance of retrieved patterns

Core Design Contradiction:

Extent of automationVSReliability

Solution Approach 1:

The patent introduces a dual-engine architecture where a retrieval engine first identifies similar cases based on quantitative parameters, and a grading engine then evaluates the clinical significance of retrieved patterns. This intermediary grading mechanism filters out spurious patterns and ensures clinical relevance, resolving the contradiction between automated analysis and reliability.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces traditional machine learning classification approaches with a similarity-based retrieval system using quantitative parameters. This substitution avoids the overfitting problems of training-based systems while maintaining automated analysis capabilities, as the system retrieves based on measured parameters rather than learned patterns from limited data.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Loss of information

If traditional machine learning classification is used, then pattern extraction from medical data is improved, but the clinical significance of extracted patterns cannot be guaranteed

Engineering Contradiction:

Improvepattern extraction capabilityVSAvoidclinical significance

Core Design Contradiction:

Loss of informationVSReliability

Solution Approach 1:

The grading engine serves as an intermediary that evaluates retrieved patterns for clinical significance. It assesses whether extracted patterns represent true disease characteristics or spurious correlations, ensuring that only clinically meaningful patterns are presented to clinicians.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent extracts specific quantitative parameters (e.g., nodule size, density, texture features) that have established clinical significance in radiology. By focusing on these clinically validated parameters rather than general pattern extraction, the system ensures clinical relevance while maintaining pattern extraction capability.

Inventive Principle:

Principle #2Taking out (Extraction)

3Productivity

If limited medical data is used for analysis, then processing time and resource requirements are reduced, but the reliability of analysis decreases due to overfitting and insufficient data

Engineering Contradiction:

Improvedata processing efficiencyVSAvoidanalysis reliability

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system creates a database of pre-processed medical images with extracted quantitative parameters that can be efficiently searched and retrieved. This copied representation allows rapid similarity searching without re-processing original images, maintaining high productivity while enabling reliable analysis through comprehensive parameter comparison.

Inventive Principle:

Principle #26Copying

4Loss of information

If quantitative parameters are extracted and stored in a database, then similar case retrieval capability is improved, but the system complexity increases

Engineering Contradiction:

Improvesimilar case retrieval accuracyVSAvoidsystem architecture complexity

Core Design Contradiction:

Loss of informationVSDevice complexity

Solution Approach 1:

The system is segmented into distinct functional modules: image preprocessing module, parameter extraction module, database storage module, retrieval engine, and grading engine. Each module handles a specific task, making the overall complex system manageable and maintainable while achieving high retrieval accuracy through specialized processing at each stage.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS12073943B2Medical image analysis system and similar case retrieval system using quantitative parameters, and methods for the same

Publication Date: 2024.08.27 CORELINE SOFT

AI summary

Disclosed herein is a computing system for performing medical image analysis. A computing system for performing medical image analysis according to an embodiment of the present invention includes at least one processor. The at least one processor performs image processing on a first medical image, and segments at least one anatomical region in the first medical image. The at least one processor generates a first quantitative parameter for the at least one anatomical region based on quantitative measurement conditions that can be measured in the first medical image, and stores the first quantitative parameter in a database in association with the first medical image and the at least one anatomical region.