Vascular Tree Reconstruction from Single 2D Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for reconstructing 3-D vascular trees from 2-D angiographic images face challenges such as noise introduced by heartbeat and respiratory movement, leading to inaccuracies in depth recovery and vessel overlap, especially when images are not taken simultaneously or at phase-locked angles.
Innovation Solution
A method that uses a single 2-D image to reconstruct 3-D positions of vascular segments by defining a surface model based on anatomical constraints, registering anchoring vascular segments, and assigning 3-D positions to associated segments, which reduces errors and circumvents the need for multiple image correspondences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple 2-D angiographic images are used for 3-D vascular reconstruction, then depth recovery accuracy improves, but noise from heartbeat and respiratory movement increases causing vessel overlap and matching errors
Solution Approach 1:
The patent extracts and removes the harmful noise components from the image processing pipeline by using a single 2-D image rather than multiple images that would introduce temporal noise variations from heartbeat and respiration. This extraction approach eliminates the source of matching errors and vessel overlap while maintaining depth recovery capability through alternative means (surface model constraints).
Solution Approach 2:
The patent transitions from temporal dimension (multiple images over time) to spatial dimension constraints (surface model geometry) for achieving depth recovery. By constraining vascular segments to lie on a defined surface model, the system recovers 3-D depth information from a single 2-D image without introducing temporal noise from multiple acquisitions.
2Manufacturing precision
If multiple 2-D images are acquired for 3-D reconstruction, then vascular structure detail improves, but procedure time increases reducing productivity
Solution Approach 1:
The patent applies preliminary action by pre-defining the surface model geometry before image acquisition. This predetermined geometric framework allows single-image 3-D reconstruction without requiring multiple sequential images, thereby maintaining vascular structure detail while dramatically reducing procedure time and improving productivity.
Solution Approach 2:
The patent creates a 3-D copy of the vascular structure from a single 2-D image by projecting vascular segments onto the pre-defined surface model. This copying process reconstructs detailed vascular geometry without requiring multiple source images, thus maintaining manufacturing precision while reducing acquisition time.
3Measurement precision
If iterative projection and image registration are performed to improve feature position matching, then 3-D vascular extent accuracy improves, but computational complexity and time increase
Solution Approach 1:
The patent performs preliminary action by pre-defining the surface model and vascular segment paths before the actual 3-D reconstruction. This pre-positioning eliminates the need for iterative projection and registration adjustments, as vascular segments are directly assigned to their 3-D positions on the surface model based on their 2-D image coordinates, thereby maintaining accuracy while reducing computational complexity.
Data Source
AI summary
Methods of and systems for reconstructing a vascular tree shape from vascular segments imaged in a single source 2-D projection image are described. A structuring shape comprising spatial positions of reference anatomical elements is defined, such as vascular segments in the definition of a 3-D surface model corresponding to a surface defined by an anatomical structure such as a body organ (e.g., heart). The 3-D surface model is used to create a 3-D model of anatomical elements (e.g., additional vascular segments of a cardiac vasculature) imaged in a source 2-D projection image, by back-projection to the 3-D surface model. The 3-D surface model is optionally aligned by first aligning the source 2-D projection image to the structuring shape. In some embodiments, the source 2-D projection image is registered to the 3-D surface model through the structuring shape by the source image's initial use in defining the structuring shape.