Patient-Specific Vascular Segmentation Without Mask Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 3D-DSA methods for cerebral vessel visualization suffer from artifacts due to patient movement, require double the recording time, additional radiation, and unsuitable data for training AI models, leading to suboptimal vessel segmentation.
Innovation Solution
A machine learning algorithm, such as an artificial neural network or support vector machine, is trained using patient-specific 3D reconstructions to segment vessels by identifying a starting vascular region, extracting subregions, and iteratively refining the algorithm through neighboring regions, utilizing vessel filters and intensity fluctuations to enhance segmentation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If 3D-DSA with mask run is used for vessel visualization, then bone interference is eliminated, but recording time doubles and additional radiation dose is required
Solution Approach 1:
The patent extracts and removes the mask run step from the traditional 3D-DSA process. By using AI-based bone suppression techniques, the method eliminates the need for separate mask images while achieving bone-free vessel visualization, thereby reducing recording time without compromising visualization quality
Solution Approach 2:
The patent changes the fundamental approach from subtractive (mask-based) to additive (AI-based bone suppression). By transforming bone suppression into a post-processing step using machine learning models, the system achieves bone elimination without requiring additional mask acquisitions, thus reducing time loss
2Reliability
If 3D-DSA with mask run is used for vessel visualization, then bone interference is eliminated, but additional x-ray radiation dose is required
Solution Approach 1:
The patent removes the mask acquisition step that causes additional radiation exposure. By replacing mask-based bone suppression with AI-based methods that work on existing filling images, the system eliminates unnecessary radiation doses while maintaining bone-free vessel visualization
Solution Approach 2:
The patent converts the limitation of having only filling images (which contain both bone and vessel information) into an advantage by using AI to selectively suppress bone signals. This transforms a potentially harmful additional radiation exposure into a beneficial radiation-efficient approach
3Quantity of substance
If standard 3D-DSA data is used for AI training, then data availability is ensured, but undersampling artifacts prevent effective training
Solution Approach 1:
The patent applies preliminary bone suppression and vessel enhancement processing to training data before AI model training. By pre-processing the 3D-DSA data to remove bone artifacts and enhance vessel signals, the system creates high-quality training datasets that enable effective AI learning without requiring additional clinical scans
Solution Approach 2:
The patent creates enhanced copies of existing 3D-DSA data through AI-based bone suppression and vessel enhancement. These processed copies serve as high-quality training datasets, allowing the system to multiply the utility of limited clinical data without additional radiation exposure or patient burden
4Measurement precision
If vessel filters are used for vessel detection, then vessel enhancement is achieved, but bones are misclassified as vessels and distal vessels are overlooked
Solution Approach 1:
The patent segments the vessel detection process into multiple specialized AI models: one for bone suppression, one for vessel enhancement, and another for accurate vessel segmentation. This multi-stage segmentation approach allows each model to specialize in specific tasks, avoiding the misclassification problems of single-filter methods
Solution Approach 2:
The patent applies different processing characteristics to different regions and vessel types. By using location-aware AI models that adapt to local anatomical features and vessel characteristics, the system achieves high detection sensitivity for distal vessels while maintaining segmentation accuracy by suppressing bone signals in bone-prone regions
Data Source
AI summary
Vessels of a biological object are to be reliably segmented. To this end, a method for training a machine learning algorithm for the purpose of segmenting such vessels is proposed. In the method, a 3D reconstruction is provided with the vessels. A starting vascular region is identified in the 3D reconstruction. Subregions that represent a starting vascular segment are extracted from the starting vascular region. The algorithm is trained with the extracted subregions. The trained algorithm is then applied to a first neighboring region, which is adjacent to the starting vascular region. As a result, a first vascular segment is determined in the first neighboring region. Finally, the algorithm is retrained with the first vascular segment determined in the first neighboring region.