DSA Image Registration via Vascular Region Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for digital subtraction angiography (DSA) image registration struggle with three-dimensional object motion, leading to incomplete registration and potential false image generation, especially in vascular regions, due to the complexity of non-rigid body registration and increased calculation costs.
Innovation Solution
An X-ray image processing technique that generates a first difference image between a mask and live image, identifies a line-shaped vascular region and its peripheral area, and performs registration by comparing positions within these regions to minimize motion artifacts, allowing for effective registration between the live and mask images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If non-rigid body registration is used to cope with three-dimensional motion, then registration accuracy is improved, but the algorithm becomes complicated and calculation cost increases greatly
Solution Approach 1:
The patent segments the image into multiple regions of interest (vascular regions, peripheral regions, bone regions) and performs registration separately for each region. This divides the complex global registration problem into simpler local registration tasks, reducing algorithmic complexity while maintaining accuracy in critical vascular areas.
Solution Approach 2:
The patent applies different registration strategies to different regions: rigid body registration for bone regions, non-rigid body registration for vascular regions, and intermediate methods for peripheral regions. This local differentiation optimizes registration accuracy where needed while reducing overall computational complexity.
2Measurement precision
If non-rigid body registration is used to cope with three-dimensional motion, then registration accuracy is improved, but calculation cost increases greatly
Solution Approach 1:
The patent segments the image into multiple regions of interest (vascular regions, peripheral regions, bone regions) and performs registration separately for each region. This divides the complex global registration problem into simpler local registration tasks, reducing algorithmic complexity while maintaining accuracy in critical vascular areas.
Solution Approach 2:
The patent applies different registration strategies to different regions: rigid body registration for bone regions, non-rigid body registration for vascular regions, and intermediate methods for peripheral regions. This local differentiation optimizes registration accuracy where needed while reducing overall computational complexity.
3Measurement precision
If registration is performed in the entire image, then complete registration is achieved, but calculation cost increases and partial optimization is lost
Solution Approach 1:
The patent segments the image into multiple regions of interest (vascular regions, peripheral regions, bone regions) and performs registration separately for each region. This divides the complex global registration problem into simpler local registration tasks, reducing algorithmic complexity while maintaining accuracy in critical vascular areas.
Solution Approach 2:
The patent performs registration selectively in specific regions rather than the entire image. By focusing computational resources on vascular regions and their peripheries where accurate registration is most critical, the method achieves sufficient registration completeness while significantly reducing calculation costs.
4Object-affected harmful factors
If motion artifact reduction is prioritized, then motion artifacts are reduced, but clinical interest regions may not be optimized
Solution Approach 1:
The patent applies different registration strategies to different regions: rigid body registration for bone regions, non-rigid body registration for vascular regions, and intermediate methods for peripheral regions. This local differentiation optimizes registration accuracy where needed while reducing overall computational complexity.
Solution Approach 2:
The patent performs registration selectively in specific regions rather than the entire image. By focusing computational resources on vascular regions and their peripheries where accurate registration is most critical, the method achieves sufficient registration completeness while significantly reducing calculation costs.
Data Source
AI summary
An X-ray image processing apparatus includes a first difference processing unit for generating a first difference image by performing difference processing between a mask image obtained by capturing an object before an inflow of a contrast medium and a live image after the inflow of the contrast medium, a first obtaining processing unit for obtaining a line-shaped region indicating a region, into which the contrast medium has flowed, using a distribution of pixel values in the first difference image, a second obtaining processing unit for obtaining a peripheral region of the line-shaped region from the first difference image using pixel values of pixels adjacent to the line-shaped region, and a registration processing unit for performing registration between pixels of the live image and the mask image by comparing positions using the line-shaped region and the peripheral region.