Automated Medical Image Quality Checks for Segmentation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The effectiveness of machine learning computer models in medical image analysis is highly dependent on the quality of training data, particularly medical images and SME annotations, leading to potential misdiagnosis or missed anomalies due to poor image quality and segmentation issues.

Innovation Solution

An automated computing tool assesses the quality of medical imaging data and trained models by generating a reference profile for anatomical structures, detecting discrepancies, and providing notifications to identify and correct errors in image acquisition and segmentation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual assessment of medical image quality and segmentation is performed, then expertise and judgment can be applied, but the process is time-consuming and resource-intensive

Engineering Contradiction:

Improveimage quality assessment accuracyVSAvoidassessment time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces manual mechanical assessment by subject matter experts with an automated computer-based system that uses machine learning models and algorithms to assess medical image quality and segmentation accuracy, thereby eliminating time-consuming manual review while maintaining assessment precision

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables self-assessment of medical image quality and segmentation through automated computational methods, allowing the system to evaluate itself without requiring external expert intervention for every image assessment

Inventive Principle:

Principle #25Self-service

2Reliability

If extensive manual review and experimentation are conducted to train machine learning models, then model effectiveness can be improved, but the process requires large amounts of manual effort and resources

Engineering Contradiction:

Improvemodel effectivenessVSAvoidtraining process complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements preliminary automated quality assessment and validation steps before full model training, pre-processing and filtering training data to ensure quality, which reduces the complexity and resource requirements of the subsequent training process while maintaining model effectiveness

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system incorporates automated feedback mechanisms where assessment results from quality control models are used to iteratively improve the main machine learning model, creating a closed-loop system that reduces manual intervention needs while enhancing model reliability

Inventive Principle:

Principle #23Feedback

3Reliability

If high quality training data with accurate segmentation is used, then machine learning model performance improves, but ensuring data quality requires significant manual effort

Engineering Contradiction:

Improvetraining data qualityVSAvoiddata preparation efficiency

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent replaces manual data quality verification and segmentation validation with automated computer-based assessment systems that use machine learning models to evaluate training data quality, thereby maintaining high data quality standards while dramatically improving preparation efficiency

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system segments the data preparation process into distinct automated quality assessment steps, separating quality control functions from manual processes and enabling parallel processing and automation of quality verification tasks

Inventive Principle:

Principle #1Segmentation

4Productivity

If automated quality assessment systems are implemented, then time and resource efficiency improve, but system complexity increases

Engineering Contradiction:

Improveassessment efficiencyVSAvoidsystem complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent implements a universal automated quality assessment system that handles multiple types of medical images and segmentation tasks through a single integrated platform, reducing overall system complexity by consolidating multiple specialized tools into one multi-functional system

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS12437406B2Automated medical image and segmentation quality assessment for machine learning tasks

Publication Date: 2025.10.07 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

Mechanisms for identifying inconsistencies between volumes of medical images are provided. A plurality of volumes of medical images are received, each having a plurality of medical images of an anatomical structure. A plurality of first representation data structures are generated, each corresponding to a volume and having dimensional measurements of the anatomical structure at various locations. A reference data structure is generated, for the anatomical structure based on the first representation data structures, having second dimensional measurements derived from the first dimensional measurements. A discrepancy is detected between a second representation data structure and the reference data structure based on a comparison and a notification of the discrepancy where the notification identifies a type of the discrepancy.