Automated X-ray Image Verification Using Segmentation and Reference Comparison

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current X-ray image quality checking methods are time-consuming, labor-intensive, and prone to errors, often requiring manual inspection by X-ray assistants, leading to unnecessary patient re-exposure and additional imaging time.

Innovation Solution

An automated method for checking X-ray image data using machine-learning algorithms for segmentation and comparison with reference data, determining the suitability of images for diagnosis and reducing the need for manual inspection.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual image review by X-ray technician is performed, then image quality assessment can be conducted, but additional time and personnel costs are required

Engineering Contradiction:
Improveimage quality assessment accuracyVSAvoidtime per patient in imaging unit
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent replaces the manual mechanical review process by X-ray technicians with an automated computer-based image analysis system. The system uses algorithmic processing to automatically assess image quality, segment anatomical structures, and determine diagnostic suitability, eliminating the need for manual mechanical inspection while maintaining or improving assessment accuracy.

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

Solution Approach 2:

The image quality assessment system performs self-evaluation through automated algorithms that independently analyze image data without human intervention. The system segments images, identifies anatomical structures, compares them against reference data, and automatically determines diagnostic quality, enabling the system to serve itself rather than requiring external technician review.

Inventive Principle:
Principle #25Self-service

2Reliability

If repeat X-ray exposure is performed to ensure image quality, then diagnostic accuracy is improved, but additional radiation dose to patient is incurred

Engineering Contradiction:
Improvediagnostic accuracyVSAvoidradiation dose to patient
Core Design Contradiction:
ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The system performs preliminary automated quality assessment of X-ray images immediately after acquisition but before clinical use. By pre-evaluating image quality through automated segmentation and analysis, the system identifies suitable images in advance, preventing the need for repeat exposures and thereby avoiding additional radiation doses to patients.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system provides immediate feedback on image quality through automated analysis, informing whether an image is suitable for diagnosis. This feedback mechanism allows clinicians to make informed decisions about image acceptance or rejection without requiring repeat exposures, thus preventing unnecessary radiation exposure while ensuring diagnostic accuracy.

Inventive Principle:
Principle #23Feedback

3Productivity

If automated image analysis is implemented, then processing speed is improved, but system complexity increases

Engineering Contradiction:
Improveimage processing speedVSAvoidautomation system complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The automated analysis system segments the complex task of image quality assessment into distinct modular components: image segmentation, anatomical structure detection, reference data comparison, and quality determination. Each module handles a specific aspect of the analysis independently, making the overall complex system more manageable and easier to implement while maintaining high processing speed.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3301642B1Automated image verification in x-ray imaging
Publication Date: 2020.07.15 SIEMENS HEALTHCARE GMBH
  • EP3301642B1 patent drawingFigure 1
  • EP3301642B1 patent drawingFigure 2~3

AI summary

A method for the automated review of radiographic image data (RI) from a patient's examination area is described. Within this method, radiographic image data (RI) from the examination area is received. Furthermore, the radiographic image data (RI) is segmented, and anatomical structures within the individual segments are detected. Additionally, the closest reference image data (RI) to the segmented radiographic image data (SRI) is determined. This process is based on a comparison of the segmented radiographic image data (SRI) with reference image data (RI) from a reference database (DB), where the reference image data (RI) in the reference database (DB) each contains quality information regarding the image quality of the reference image data (RI). Finally, based on the quality information of the determined reference image data (RI), a decision is made as to whether the received radiographic image data (RI) is retained or discarded.An imaging procedure is also described. Furthermore, an image data verification device (20) is described. Additionally, an X-ray imaging device (30) is described.