Deformable 2D-3D Vascular Registration via Graph Energy Minimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for 2D-3D registration in vascular interventions, such as AAA repair, face challenges in accurately aligning pre-operative 3D volumetric data with intra-operative 2D X-ray images due to the complex and deformable nature of vascular structures, which existing approaches struggle to correct for local deformations effectively.
Innovation Solution
A method for deformable non-rigid registration using graph-based segmentation and energy minimization, where the vascular structure is represented as a graph, and the difference between 3D and 2D image data is expressed as a sum of distance, length preservation, and deformation smoothness energies, allowing for accurate alignment despite changes in the vascular anatomy during procedures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If rigid alignment methods are used for 2D-3D registration, then the general pose estimation is robust, but local deformations within the vessel cannot be effectively corrected
Solution Approach 1:
The patent segments the vascular structure into multiple control points or regions along the vessel centerline. Each segment can be independently deformed to match the 2D projection, allowing local deformation correction while maintaining overall pose accuracy. The graph-based representation divides the vessel into discrete nodes that can be individually adjusted.
Solution Approach 2:
The patent transitions from rigid transformation to deformable transformation by introducing degree-of-freedom parameters for each control point. The system dynamically adjusts the position and shape of vessel segments based on the energy minimization process, enabling the model to adapt to local deformations caused by body motion, heart beat, and breathing.
2Manufacturing precision
If deformable 2D/3D registration methods are used, then local deformations can be corrected, but the methods are ill-suited for complex and repetitive vascular structures with high degree of deformation
Solution Approach 1:
The patent applies different properties to different parts of the vascular structure by assigning specific constraints and energy weights to different vessel segments. The graph-based model allows each node and edge to have localized characteristics, enabling the system to handle complex and repetitive structures by treating each segment with appropriate local quality parameters.
Solution Approach 2:
The patent uses energy minimization with adjustable parameters (weights for different energy terms, control point spacing, deformation constraints) to adapt the registration method to various vascular structures. By changing these parameters, the system can handle different degrees of deformation and various vascular geometries effectively.
3Loss of information
If pre-operative 3D volumetric data is combined with intra-operative 2D X-ray images, then realistic artery anatomy can be visualized with minimal radiocontrast, but the size, shape and relative location of vascular anatomy change due to body motion, heart beat, breathing and device insertion
Solution Approach 1:
The patent models the vascular anatomy as a dynamic structure that changes over time due to body motion, heart beat, and breathing. The deformable registration framework continuously adjusts the 3D model to match the current 2D projection, maintaining alignment despite anatomical changes. The system captures temporal variations by allowing the vessel centerline and cross-sections to deform dynamically.
Solution Approach 2:
The patent uses feedback from the 2D X-ray images to continuously correct the 3D model during the procedure. The energy minimization process compares the projected 3D model with the actual 2D image and adjusts the model parameters accordingly, providing real-time feedback that compensates for anatomical changes caused by device insertion and physiological movements.
Data Source
AI summary
A method for performing deformable non-rigid registration of 2D and 3D images of a vascular structure for assistance in surgical intervention includes acquiring 3D image data. An abdominal aorta is segmented from the 3D image data using graph-cut based segmentation to produce a segmentation mask. Centerlines are generated from the segmentation mask using a sequential topological thinning process. 3D graphs are generated from the centerlines. 2D image data is acquired. The 2D image data is segmented to produce a distance map. An energy function is defined based on the 3D graphs and the distance map. The energy function is minimized to perform non-rigid registration between the 3D image data and the 2D image data. The registration may be optimized.