4D Flow MRI Vessel Segmentation Using SDM Velocity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 4D flow MRI techniques for vessel segmentation in cardiovascular applications suffer from acquisition and processing errors, leading to inaccurate hemodynamic metric calculations, particularly in tortuous cerebral vessels, and existing automated segmentation methods are not generalizable or robust to variations in MR scan parameters and flow types.
Innovation Solution
The use of a standardized difference of means (SDM) velocity metric to identify voxels with significant flow effects by quantifying the ratio between net flow and observed pulsatility, combined with an iterative segmentation algorithm that includes p-value estimation and post-processing steps to refine vessel segmentation, addressing errors and inconsistencies in 4D flow MRI data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated segmentation methods using signal magnitude images are used, then segmentation speed is improved, but segmentation accuracy deteriorates due to variability in signal magnitude throughout the field-of-view
Solution Approach 1:
The patent changes the parameter used for segmentation from signal magnitude to velocity data. Specifically, it uses the pseudo-complex difference (PCD) method which processes velocity information rather than magnitude information, thereby eliminating the field-of-view variability issue while maintaining automated segmentation capabilities
Solution Approach 2:
The patent introduces an intermediary processing step that converts velocity data into a segmentation-friendly format. The PCD method acts as an intermediary that transforms the velocity field into a form that can be directly thresholded for segmentation, bridging the gap between velocity measurement and accurate segmentation
2Measurement precision
If deep learning approaches are used for vessel segmentation, then segmentation accuracy is improved, but adaptability to different vascular systems and MR sequences deteriorates
Solution Approach 1:
The patent divides the segmentation problem into simple, independent steps: velocity calculation, PCD transformation, and thresholding. This segmented approach makes the method more adaptable to different vascular systems and MR sequences compared to monolithic deep learning models, as each step can be independently adjusted for different applications
Solution Approach 2:
The patent uses parameter changes in the PCD method to adapt to different vascular systems. By adjusting the velocity threshold and other parameters, the same basic algorithm can be applied to various vascular beds (cerebral, cardiac, peripheral) and different MR sequences without requiring retraining of deep learning models
3Measurement precision
If velocity-based segmentation methods are used, then sensitivity to flow effects is improved, but reliability deteriorates due to inapplicability to steady flow patterns
Solution Approach 1:
The patent changes the temporal parameter handling by using the pseudo-complex difference method which compares velocity at different time points. This allows the method to detect steady flow patterns by identifying consistent velocity directions across time, rather than requiring pulsatile flow variations
Solution Approach 2:
Instead of using pulsatility as the basis for segmentation (which fails for steady flow), the patent inverts the approach by using the consistency and directionality of velocity vectors across time. This inversion allows steady flow patterns to be reliably detected and segmented
Data Source
AI summary
The invention generally provides systems and methods for performing vessel segmentation from flow data, such as but not limited to 4D flow Magnetic Resonance Imaging (MRI) data. In certain aspects, the systems and methods of the invention may involve receiving flow data representative of flow in a vessel (such as 4D MRI flow data); identifying net flow effects in the flow data (such as 4D MRI flow data) according to a standardized difference of means (SDM) velocity that involves quantifying a ratio between net flow and observed flow pulsatility in each voxel of the received flow data (such as 4D MRI flow data); and identifying voxels with higher SDM velocity values than stationary tissue voxels, thereby performing vessel segmentation from flow data (such as 4D MRI flow data).