Model Regularized Motion Compensation for Medical Image Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Motion artifacts such as blurring and streaks in medical imaging, particularly in CT reconstruction, are not effectively removed by existing methods, especially when caused by external objects like electrodes during cardiac or respiratory motion, leading to inaccurate motion vector calculations.
Innovation Solution
The implementation of model regularized motion compensated reconstruction, where anatomical structures are segmented using fitted models, and volumetric masks are constructed to exclude external artifacts, allowing for accurate motion estimation and compensation in image reconstruction, enhancing vascular structures with filters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If elastic registration is used between entire volumes of each reconstructed multi-phase image, then motion compensation is achieved, but motion artifacts from external objects (electrodes) cause inaccurate motion vector calculations
Solution Approach 1:
The patent segments the image volume into multiple regions: a first region containing the object of interest (e.g., heart), a second region containing external objects (e.g., electrodes), and a third region containing intermediate structures. This segmentation allows motion compensation to be applied selectively to the first region while excluding the second region, preventing artifacts from external objects from corrupting motion vector calculations.
Solution Approach 2:
The patent extracts and excludes the second region (containing external objects like electrodes) from the motion compensation process. By identifying and removing these artifact-generating regions from the registration calculation, the system prevents them from introducing inaccurate motion vectors while preserving motion compensation for the anatomical structures of interest.
2Reliability
If motion compensation is applied to entire image volumes, then all motion is compensated, but computational complexity increases and artifacts from external objects are included
Solution Approach 1:
The patent divides the image volume into distinct regions (first region with anatomical structures, second region with external objects, third region with intermediates) and applies motion compensation only to the first region. This selective approach reduces computational complexity compared to processing entire volumes while maintaining image quality for the regions that require compensation.
Solution Approach 2:
The patent applies different processing treatments to different regions: motion compensation is applied to the first region (anatomical structures), while the second region (external objects) is excluded from compensation. This local differentiation optimizes computational resources by focusing processing only where needed while avoiding artifacts from regions where compensation would be harmful.
3Loss of information
If multi-phase reconstruction is performed, then motion information is captured, but streak artifacts from electrodes vary with angular position and mimic motion
Solution Approach 1:
The patent segments the reconstructed volume into regions, identifying the second region as containing external objects that generate streak artifacts. By excluding this region from motion analysis, the system prevents artifact-induced false motion signals from contaminating the motion information captured from genuine anatomical structures in the first region.
Solution Approach 2:
The patent introduces a region identification and exclusion mechanism that acts as an intermediary between the raw multi-phase reconstruction data and the motion compensation process. This intermediary step identifies and isolates artifact-generating regions, allowing genuine motion information from anatomical structures to be extracted without contamination from streak artifacts caused by external objects.
Data Source
AI summary
A medical imaging system (200) includes a masking unit (234), an image registration unit (238), a motion estimator (240) and a motion compensating reconstructor (244). The masking unit constructs a mask for each reconstructed volumetric phase image of a plurality of reconstructed volumetric phase images that masks portions of a corresponding image external to an anatomical model fitted to a segmented at least one anatomical structure, 5 wherein the plurality of reconstructed volumetric phase images include a target phase and a plurality of temporal neighboring phases reconstructed from projection data. The image registration unit registers the masked reconstructed volumetric phase images. The motion estimator estimates motion between the target phase and the plurality of temporal neighboring phases according to the model based on the registered masked reconstructed 10 volumetric phase images. The motion compensating reconstructor reconstructs a motion compensated medical image from the projection data using the estimated motion of the registered masked reconstructed volumetric phase images.