Elastic Image Registration Using Segmented Affine and Elastic Transformations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current diagnostic imaging techniques face challenges in accurately aligning images with non-rigid motions, requiring manual correction of 3D image registration due to the complexity of elastic transformations, which is difficult and time-consuming, especially when dealing with large volumes of data.
Innovation Solution
An automated registration system that uses a combination of affine and elastic transformations to align images, followed by real-time manual correction tools like Gaussian pull and push tools for precise alignment of 2D slices, allowing for efficient correction and display of aligned images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If elastic transformation is used to register images with non-rigid motions, then the alignment accuracy is improved, but the computational complexity and time required increase significantly
Solution Approach 1:
The patent segments the image registration process into two distinct stages: rigid transformation (for global alignment) and elastic transformation (for local non-rigid alignment). This segmentation allows each stage to handle specific aspects of alignment independently, reducing overall computational complexity while maintaining accuracy for both rigid and non-rigid motions.
Solution Approach 2:
The patent applies rigid transformation as a preliminary step before elastic transformation. By first establishing global alignment through rigid transformation, the subsequent elastic transformation only needs to handle local deviations, significantly reducing the computational burden and time required for the more complex elastic registration.
2Measurement precision
If manual correction of 3D image registration is performed, then the alignment precision is improved, but the time consumption and operational difficulty increase
Solution Approach 1:
The patent extracts and displays only the 2D slices that require manual correction from the full 3D dataset. By taking out only the necessary 2D information for user interaction, the system reduces the amount of data the user must process while maintaining the ability to perform precise manual corrections on the critical planes.
Solution Approach 2:
The patent transforms the manual correction task from 3D volume editing to 2D slice editing. By working with 2D slices instead of manipulating entire 3D volumes, the user can perform corrections more efficiently with less time consumption, while the system maintains the full 3D registration context.
3Measurement precision
If 3D datasets are manually deformed for registration correction, then the registration accuracy is improved, but the difficulty of operation and data volume increase
Solution Approach 1:
The patent extracts only the 2D slice data for manual deformation operations, separating the correction task from the full 3D dataset. This extraction reduces the operational complexity and makes manual correction more manageable, while still achieving accurate registration through the corrected 2D slices.
Solution Approach 2:
The patent shifts the operational dimension from 3D volume deformation to 2D slice deformation. By restricting manual operations to 2D slices, the system significantly improves ease of operation while maintaining registration accuracy, as users can intuitively manipulate 2D images rather than complex 3D volumes.
Data Source
AI summary
A current diagnostic image (A) and an archived diagnostic image (B) of a common region of a patient are loaded into a first memory and a second memory. The first and second diagnostic images (A, B) are automatically aligned and registered with one another. Three 2D orthogonal views through a selected crossing point in the current image (A) are concurrently displayed along with the same three orthogonal views through the corresponding crossing point in the archived image (B) on a display. A user manually corrects alignment in the first and second sets of slices that are currently displayed on the display using local tools.