Multi-View 2D Projection Segmentation With Inter-View Consistency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing neural networks (NNs) for processing 2-D projection images exhibit limited accuracy in detecting and classifying regions of interest (ROIs) due to the projective nature of these images, which can lead to false positives and inconsistencies when viewed from different angles.
Innovation Solution
A method and system that utilize a neural network algorithm to determine inter-view consistency between multiple 2-D projection images through predefined registration, enabling accurate 2-D and 3-D segmentations of ROIs by enforcing segmentations to align with a reference frame, thereby improving detection and classification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple 2-D projection images from different views are used to improve detection accuracy, then the ability to classify pathologies improves, but the complexity of processing and ensuring consistency between views increases
Solution Approach 1:
The patent segments the processing task by first determining 2-D segmentations for each individual view independently, then separately determining inter-view consistency between these segmentations. This divides the complex multi-view processing into manageable stages: individual view processing followed by consistency verification, reducing overall processing complexity while maintaining detection accuracy.
Solution Approach 2:
The patent introduces 2-D segmentations as an intermediary representation that bridges individual view processing and multi-view consistency analysis. By working with segmentations rather than raw images directly, the system creates an intermediate layer that simplifies the comparison and consistency determination across multiple views.
2Extent of automation
If neural network algorithms are used to detect ROIs in 2-D projection images, then detection automation improves, but the accuracy is limited due to the projective nature of the images
Solution Approach 1:
The patent implements a feedback mechanism where the determined 2-D segmentations from multiple views are used to compute inter-view consistency, which then feeds back to refine and validate the ROI detections. This feedback loop allows the automated system to self-correct inconsistencies arising from the projective nature of individual views, thereby improving detection accuracy while maintaining automation.
Solution Approach 2:
The patent merges information from multiple independent 2-D segmentations by determining their inter-view consistency. By combining the segmentations from different views and verifying their consistency, the system overcomes the limitations of individual projective views and achieves higher detection accuracy through the synergistic integration of multiple perspectives.
3Loss of information
If 2-D segmentations are determined for multiple views, then the information available for classification improves, but ensuring consistency between segmentations from different views becomes more difficult
Solution Approach 1:
The patent performs preliminary determination of 2-D segmentations for each view before addressing consistency issues. By first obtaining segmentations from all views and then systematically determining their inter-view consistency, the system ensures that no view is overlooked while maintaining reliability through structured consistency verification.
Data Source
AI summary
Techniques are described to infer 2-D segmentations of a region of interest using a neural network algorithm. Techniques are described to train the neural network algorithm. The 2-D segmentations are determined based on multiple 2-D projection images. For example, x-ray images can be used as an input.