CT Image Reconstruction via Frequency Domain Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cone-beam artifacts degrade the quality of reconstructed CT images, especially in larger coverage CT scanners, due to data truncation, uneven sampling, and incorrect weighting, leading to shading and glaring around high contrast edges, and introducing new artifacts.
Innovation Solution
A method involving different view weightings for intermediate reconstructions, transforming data into frequency space, masking to select relevant frequency information, and combining these to generate a fully sampled frequency space for improved image reconstruction, reducing cone-beam artifacts and computational intensity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional cone-beam reconstruction techniques are used, then reconstruction can be performed, but cone-beam artifacts are introduced that degrade image quality
Solution Approach 1:
The image space is divided into multiple subdivisions, and the frequency space is segmented into different regions. Each subdivision is processed separately with appropriate masking to select only the relevant frequency information for that region, avoiding the propagation of artifacts from one region to another.
Solution Approach 2:
The patent transforms the reconstruction problem from spatial domain to frequency domain using Fourier transforms. By operating in frequency space and applying masks to select specific frequency components, the method eliminates cone-beam artifacts that are problematic in the spatial domain.
2Reliability
If conventional reconstruction techniques are used, then image reconstruction is achieved, but computational cost and processing time are high
Solution Approach 1:
The reconstruction process is segmented into discrete steps: acquiring projection data, performing initial reconstructions, transforming to frequency space, applying masks, combining frequency information, and inverse transforming. This segmentation allows for optimized computation at each stage and avoids unnecessary calculations.
Solution Approach 2:
The patent changes the domain of operation from spatial to frequency domain, which fundamentally alters the computational characteristics of the reconstruction process. This parameter change enables more efficient computation by operating on frequency components rather than spatial pixels.
3Ease of operation
If data is acquired along circular or circular segment trajectories, then scanning is simplified, but cone-beam artifacts are produced due to incomplete sampling
Solution Approach 1:
By transforming to frequency space, the patent can properly handle and combine frequency information from circular or circular segment trajectories. The frequency domain approach allows for correct reconstruction even when spatial sampling is incomplete, as long as the frequency coverage is sufficient.
Solution Approach 2:
The method changes the weighting parameters applied to different views and regions of the data. By adjusting view weightings and applying frequency masks, the reconstruction compensates for the incomplete sampling inherent in circular trajectories, maintaining image quality while preserving scanning simplicity.
Data Source
AI summary
Approaches for performing computed tomographic image reconstruction are described. In one embodiment, a full or almost full scan of scan data is acquired and a plurality of image reconstructions are performed based on the scan data, wherein the plurality of image reconstructions result in a corresponding plurality of image volumes wherein the image reconstructions use different view weighting functions. Further, the present approaches provide for combining the plurality of image volumes together to produce a final image volume.