Graph-Based Aggregation for Automated Medical Image Sequence Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated systems for medical image analysis, particularly in angiography, fail to effectively evaluate entire image sequences due to the absence of interrelationships between frames, limiting their acceptance and reliance on manual expert analysis.
Innovation Solution
A method and system utilizing graph-based aggregation to determine diagnostic candidates across multiple frames, generating a graph with nodes representing diagnostic candidates and edges indicating overlaps, and forming communities to identify key frames and consolidate diagnostic findings, thereby enabling automated evaluation of image sequences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated detection is applied to individual frames using machine learning algorithms, then detection speed and productivity are improved, but the ability to evaluate entire image sequences with interrelationships is lost, worsening measurement precision and reliability
Solution Approach 1:
The patent segments the image sequence evaluation into two levels: individual frame analysis using machine learning algorithms for rapid detection, and sequence-level aggregation using graph theory to establish interrelationships. This segmentation allows both automated speed and contextual accuracy to coexist by processing frames independently then integrating results through graph-based community detection.
Solution Approach 2:
The patent merges machine learning detection results with graph theory aggregation to create a comprehensive evaluation system. The graph structure combines diagnostic candidates from multiple frames, and community detection algorithms integrate these candidates into coherent diagnostic findings, merging the strengths of both approaches.
2Measurement precision
If manual expert analysis is used to evaluate entire image sequences, then measurement precision and reliability are improved, but productivity and time consumption are worsened
Solution Approach 1:
The patent introduces graph theory and community detection algorithms as intermediary systems between manual expert analysis and raw image data. These intermediaries automatically perform the complex task of integrating information across frames and identifying diagnostic patterns, achieving expert-level accuracy without requiring actual expert time investment.
Solution Approach 2:
The system performs preliminary automated detection and candidate identification across all frames before final diagnostic aggregation. This preliminary action filters and prepares data in advance, reducing the complexity of subsequent integration tasks and enabling rapid comprehensive evaluation.
3Measurement precision
If graph-based aggregation is applied to integrate diagnostic candidates across multiple frames, then measurement precision and reliability are improved, but device complexity and computational requirements are worsened
Solution Approach 1:
The patent employs graph theory as a universal framework that can handle multiple diagnostic candidates, multiple frames, and various relationship types within a single unified system. This multi-functional approach manages complexity by providing a single versatile mathematical structure rather than separate processing systems for each diagnostic task.
Data Source
AI summary
One or more example embodiments of the present invention relates to a method for the automated determination of examination results in an image sequence from multiple chronologically consecutive frames, the method comprising determining diagnostic candidates in the form of contiguous image regions in the individual frames for a predefined diagnostic finding; and for a number of the diagnostic candidates, determining which candidate image regions in other frames correspond to the particular diagnostic candidate, determining whether the candidate image regions of the particular diagnostic candidate in the other frames overlap with other diagnostic candidates, generating a graph containing the determined diagnostic candidates of the frames as nodes and the determined overlaps as edges, and generating communities from nodes connected via edges.