Vessel Graph Aggregation for Motion-Consistent Structure Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing medical imaging technologies face challenges in generating a consistent representation of vessel structures across multiple frames due to inconsistencies caused by patient motion, cardiac motion, and varying angulation, leading to reduced reliability and accuracy in subsequent tasks.
Innovation Solution
A computer-implemented method that generates an aggregated representation of a vessel structure by attaching segments of a first graph representation to a second graph representation, augmenting the second graph representation if necessary, to ensure all segments are included, using spatial coordinates for alignment and potentially employing machine learning models for registration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If graph representations are extracted from multiple frames to improve vessel structure understanding, then the completeness of vessel information is improved, but inconsistencies arise due to motion and varying angulation
Solution Approach 1:
The patent merges multiple graph representations extracted from different frames by attaching segments from a first graph representation to a second graph representation. This combining approach consolidates vessel structure information across frames while using spatial coordinates and machine learning models to ensure consistent integration, thereby reducing information loss without compromising representation reliability.
2Measurement precision
If segments are attached from one graph representation to another to ensure completeness, then the accuracy of vessel structure representation is improved, but the complexity of the aggregation process increases
Solution Approach 1:
The patent replaces manual or rule-based segment attachment with machine learning models that automatically determine which segments to attach and how to integrate them. This substitution of mechanical/manual processes with intelligent algorithms improves vessel structure accuracy while managing aggregation complexity through automated decision-making.
Solution Approach 2:
The patent utilizes spatial coordinates as key parameters to guide the segment attachment process. By changing and comparing spatial coordinate parameters across different graph representations, the system accurately identifies corresponding segments and attaches them appropriately, improving representation accuracy while maintaining a systematic aggregation approach.
3Reliability
If all segments from multiple frames are aggregated to improve diagnostic accuracy, then the reliability of downstream tasks is improved, but computational resources and processing time increase
Solution Approach 1:
The patent extracts only the necessary vessel structure segments from multiple frames rather than processing all data uniformly. By identifying and extracting relevant segments that contribute to vessel structure completeness, the system improves downstream task reliability while reducing unnecessary computational overhead and processing time.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
Graph representations (11) of a vessel structure (18) are received, each corresponding to a medical image (10) and comprising a plurality of nodes and respective spatial coordinates for each node, wherein the plurality of nodes is arranged according to at least one segment (12a, 12b). For each segment (12a, 12b) of a first graph representation (11), it is determined whether a second graph representation (11) of the plurality of graph representations (11) comprises a segment (12a, 12b) corresponding to the respective segment (12a, 12b) of the first graph representation (11) and, if it is found that this is not the case, the second graph representation (11) is augmented by adding an additional segment (12a, 12b) corresponding to the respective segment (12a, 12b) of the first graph representation (11). An aggregated representation of the vessel structure (18) is generated depending on the augmented second graph representation (11).