Medical Image Registration via Local Force Fields
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional medical image processing methods using global potential fields for non-rigid registration are prone to numerical instability and minimum solution issues, especially when handling noisy medical images and volume data, such as those from the heart in different phases.
Innovation Solution
A medical image processing apparatus and method that divides the voxel space into local force fields, calculates local forces based on node positions and image data, and deforms nodes within these fields to achieve accurate non-rigid registration, using a processor to iteratively calculate and apply local forces for precise image matching.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If global potential fields are used for non-rigid registration, then image matching can be performed, but numerical instability and minimum solution issues occur
Solution Approach 1:
The patent divides the global potential field into multiple local force fields, each responsible for a specific region. This segmentation allows the system to avoid numerical instability in global optimization while maintaining registration accuracy through localized force calculations that converge more reliably.
Solution Approach 2:
The patent applies different force field characteristics to different regions by calculating local forces based on regional image gradients and features. This local quality approach enables adaptive registration that maintains precision in critical areas while avoiding numerical issues in other regions.
2Measurement precision
If local force fields are used for non-rigid registration, then numerical stability and accuracy are improved, but computational complexity increases
Solution Approach 1:
By segmenting the computational domain into local force fields, the patent reduces the complexity of each individual calculation while maintaining overall accuracy. Each local force field operates independently with smaller computational requirements, avoiding the need for complex global optimization.
Solution Approach 2:
The patent combines multiple local force field results to achieve the final registration transformation. This merging approach allows parallel computation of local fields and simplifies the overall computational process compared to solving a single complex global problem.
Data Source
AI summary
There is provided a medical image processing apparatus matching a plurality of image data. The medical image processing apparatus includes: an image data storage that stores at least two image data of different phases of single target object; a node creating portion that creates nodes, wherein the nodes are related to positions in each of the at least two image data; a local force field calculating portion that calculates local force fields for the nodes, based on positions of the nodes and the at least two image data; a local force calculating portion that calculates local forces, each of which is acted in a corresponding one of the local force fields, from the local force fields; and an image deforming portion that deforms the positions of the nodes based on the local forces.