Image Registration Deformation Field Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current medical image registration techniques face challenges in accurately matching corresponding points between images captured by different modalities, leading to insufficient registration accuracy due to constraints other than the degree of coincidence of corresponding points.
Innovation Solution
An image processing apparatus that combines first and second deformations based on distances from corresponding points, using a corresponding point obtaining unit, first deformation obtaining unit, and second deformation obtaining unit to generate a combined displacement field, ensuring accurate matching of corresponding points while adhering to desired constraint conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a regularization term is added to the cost function to suppress physically unnatural deformation, then the overall registration quality improves, but the registration accuracy of corresponding point positions deteriorates
Solution Approach 1:
The patent applies local quality by using different weighting coefficients for different spatial locations. Specifically, the weighting coefficient is set to be larger for positions closer to corresponding points and smaller for positions farther away. This allows the cost function to locally prioritize corresponding point matching accuracy near the points while maintaining overall registration quality through the regularization term in regions farther from corresponding points.
2Measurement precision
If corresponding points are visually identified by users and used for registration processing, then registration accuracy at specific points improves, but the overall registration naturalness deteriorates due to physically unnatural deformation
Solution Approach 1:
The patent changes the parameter weighting coefficients dynamically based on spatial position. The weighting coefficient for the corresponding point coincidence term is set to decrease with distance from corresponding points, while the weighting coefficient for the regularization term increases with distance. This parameter variation allows the system to prioritize point matching accuracy locally while ensuring globally natural deformation.
Data Source
AI summary
An image processing apparatus comprises a corresponding point obtaining unit configured to obtain corresponding points that are feature points of an object associated between a first image and a second image; a first deformation obtaining unit configured to obtain first deformation that is based on the corresponding points; a second deformation obtaining unit configured to obtain second deformation that is performed between the first image and the second image, and in which a degree of coincidence of corresponding points is lower than the first deformation; and a combining unit configured to combine the first deformation and the second deformation, regarding positions in at least one image of the first Image and the second image, based on distances between the corresponding points and the positions.