Cone-Beam CT Artifact Reduction via Z-Axis Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Images reconstructed from 3D volumetric data in CT scans suffer from artifacts like 'splay' or 'windmill' artifacts due to under-sampling in the z-axis direction, which are difficult to eliminate without causing blurring of image resolution or requiring costly upgrades to the x-ray tube.
Innovation Solution
A method that processes 3D volumetric image data along the z-axis to remove high-frequency components corresponding to structured artifacts, using a de-noiser processor and subtractor to isolate and subtract noise while preserving high-frequency components that do not correspond to artifacts, thereby reducing structured artifacts without compromising image resolution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If thicker slices are reconstructed to meet Nyquist sampling criteria, then structured artifacts are reduced, but z-resolution is blurred
Solution Approach 1:
The artifact reduction process is segmented into distinct processing stages: first processing coronal slices to reduce horizontal artifacts, then sagittal slices for vertical artifacts, and finally axial slices for remaining artifacts. Each stage targets specific artifact orientations while preserving image resolution through selective frequency filtering in different anatomical planes.
Solution Approach 2:
Different filtering strategies are applied to different regions and orientations of the image data. Coronal processing addresses artifacts in one orientation, sagittal processing addresses another orientation, and axial processing handles remaining artifacts. This localized approach allows artifact reduction without uniform blurring across the entire z-resolution.
2Measurement precision
If focal spot is shifted in z-direction to stagger detector positions, then sampling is improved, but device complexity and cost increase
Solution Approach 1:
The patent replaces the mechanical approach of physically shifting the focal spot with a computational approach. Instead of modifying the x-ray tube hardware to implement focal spot staggering, the invention uses software-based processing of coronal, sagittal, and axial slices to achieve the same sampling improvement effect, thereby avoiding increased device complexity and cost.
3Reliability
If adaptive upsampling is used to smooth gradients, then artifacts are reduced, but fine structures are blurred
Solution Approach 1:
Rather than applying uniform adaptive upsampling across the entire volume, the patent segments the processing into three orthogonal planes (coronal, sagittal, axial). Each plane is processed independently with filtering tailored to the artifact patterns in that orientation, allowing artifact reduction while preserving fine structures that would otherwise be blurred by global smoothing.
Solution Approach 2:
The patent applies filtering selectively to remove only the artifact components at specific frequencies in each plane, rather than applying excessive smoothing across all frequencies. This partial action approach targets only the harmful artifact frequencies while leaving the frequency components corresponding to fine structures intact.
Data Source
AI summary
A method includes reducing structured artifacts in 3D volumetric image data, which is generated with reconstructed projection data produced by an imaging system (100), by processing the 3D volumetric image data along a z-axis (108) direction. The 3D volumetric image data includes structured artifacts which have high-frequency components in the z-axis direction, and lower-frequency compounds within the x-y plane.