Brain Image Registration Using Modified Mutual Information
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image registration techniques for brain images, particularly in deep brain surgery, face challenges in accurately aligning MRI and CT images to localize electrodes and assess surgical success, as they lack efficient methods for precise spatial alignment and fusion of multi-modal images.
Innovation Solution
A computer-implemented method and system for registering brain images using modified mutual information, which involves a coarse registration module for initial alignment based on mass sharing in predetermined directions, followed by fine registration through modified mutual information normalization, overlap rate, and weight assignment, enabling precise rotation and translation of images in axial, sagittal, and coronal directions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current image registration techniques are used for brain images, then the registration process can be completed, but the accuracy of spatial alignment between MRI and CT images is insufficient
Solution Approach 1:
The patent applies preliminary action by performing coarse registration first to achieve initial spatial alignment between MRI and CT images, then proceeding to fine registration. This two-stage approach prepares the images for more accurate electrode localization by establishing a preliminary aligned framework before applying the modified mutual information algorithm for precise registration.
Solution Approach 2:
The patent changes the parameter of mutual information by introducing a modified version that incorporates overlap rate and weight assignment. This parameter modification enhances the registration algorithm's ability to achieve accurate spatial alignment, directly improving both measurement precision and reliability in electrode localization.
2Adaptability or versatility
If traditional registration methods are used, then the process is simpler, but the ability to fuse multi-modal images effectively is reduced
Solution Approach 1:
The patent segments the registration process into distinct modules: coarse registration module and fine registration module. This segmentation allows the system to handle different aspects of multi-modal image fusion separately, improving adaptability while managing complexity through modular design. Each module can be optimized independently for its specific function.
Solution Approach 2:
The patent introduces modified mutual information as an intermediary mechanism that bridges MRI and CT images during the fine registration stage. This intermediary algorithm enables effective fusion of multi-modal images by computing similarity measures that work across different imaging modalities, enhancing versatility without requiring complete redesign of the registration system.
3Productivity
If coarse registration is performed first, then initial alignment is achieved, but fine registration requires additional computational steps
Solution Approach 1:
The patent applies preliminary action by performing coarse registration first to achieve initial spatial alignment between MRI and CT images, then proceeding to fine registration. This two-stage approach prepares the images for more accurate electrode localization by establishing a preliminary aligned framework before applying the modified mutual information algorithm for precise registration.
Solution Approach 2:
The patent employs dynamics by adapting the registration approach based on the stage of processing: using mass sharing for coarse alignment and modified mutual information for fine alignment. This dynamic switching of algorithms optimizes both productivity and precision, as each algorithm is applied where it is most effective in the registration workflow.
Data Source
AI summary
A method and a system for registering multi-modal brain images includes registering two or more full volumes of brain images based on a projection based reference image. The registration is based on sharing of mass in a predetermined direction such as axial, sagittal and coronal direction. After the registration of two or more brain images, a two or more partial volumes in a region of interest is registered through a modified mutual information. The modified mutual information may be a factor of an overlap rate and a weight assigned to it.