Medical Image Annotation Validation Workflow

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Automated image annotation systems in medical imaging lack effective validation and correction mechanisms, which are crucial for ensuring the accuracy and reliability of patient care decisions, as they often rely on complex and precise image segmentation and annotation processes.

Innovation Solution

A computer-implemented method and system that guides users through a workflow to validate and correct automated image annotations by presenting tasks in a dependent order, allowing users to review and modify annotations interactively, with tools like Smart Paint and Nudge for segmentation refinement, and a red flags task to assess image appropriateness.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If automated image annotation is used to analyze medical images, then productivity is improved, but reliability deteriorates due to lack of validation mechanisms

Engineering Contradiction:
Improveimage analysis efficiencyVSAvoidannotation accuracy
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The system implements a feedback mechanism where automated annotation results are presented to users for validation and correction. Users can review annotations, provide feedback on accuracy, and correct errors, which then feeds back into improving the automated system's future performance while maintaining high productivity

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system enables users to independently validate and correct annotations using provided tools without requiring expert intervention for every case. The smart paint and nudge tools allow users to self-correct annotations, maintaining reliability while preserving automated workflow efficiency

Inventive Principle:
Principle #25Self-service

2Reliability

If comprehensive validation and correction mechanisms are implemented, then reliability is improved, but device complexity increases

Engineering Contradiction:
Improveannotation accuracyVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system introduces an intermediary layer between automated annotation and final use - a validation interface with simplified tools that mediates the interaction. This intermediary provides comprehensive validation capabilities without exposing users to the full complexity of the underlying annotation system

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system extracts and separates the validation and correction functions from the core automated annotation process. By taking out these functions into a dedicated validation interface with specialized tools, the system improves reliability without complicating the main automated workflow

Inventive Principle:
Principle #2Taking out (Extraction)

3Manufacturing precision

If users review and correct annotations interactively, then manufacturing precision is improved, but loss of time increases

Engineering Contradiction:
Improveannotation precisionVSAvoidprocessing time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The system provides disposable, easy-to-use correction tools (smart paint, nudge) that require minimal effort and time per correction. These simple tools allow rapid precision adjustments without requiring complex, time-consuming editing processes

Inventive Principle:
Principle #27Cheap short-living objects (Disposable)

Solution Approach 2:

The system allows users to perform partial validation - reviewing only critical annotations or areas of concern rather than every annotation. This partial action approach maintains sufficient precision for clinical decision-making while reducing overall processing time

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3846176B1Systems and methods for validating and correcting automated medical image annotations
Publication Date: 2024.09.04 HEARTFLOW INC
  • EP3846176B1 patent drawingFigure 1A
  • EP3846176B1 patent drawingFigure 1B
  • EP3846176B1 patent drawingFigure 1C

AI summary

Systems and methods are disclosed for manipulating image annotations. One method includes receiving (111) an image of at least a portion of an individual's anatomy, automatically determining (113) a set of annotations for anatomical features identified in the image, determining (115) a dependency or hierarchy between at least a first and second annotation in the set, presenting (117) the set of annotations to a user for validation, including presenting the first and second annotation in an order defined by the dependency/hierarchy, determining one or more annotations as having critical status based on aspects of the annotations including one or more of: a size, shape, appearance or density of the annotation, or a relationship of the annotation to other annotations among the set, and re-presenting (119) the annotations having critical status to the user for validation or presenting the annotations having critical status to another user for validation.