MPR Streak Artifact Reduction via 3D Volume Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multiplanar reconstruction (MPR) images in computed tomography (CT) often suffer from linear streak artifacts due to cone beam artifacts, poor sampling in the z-direction, beam hardening, and mis-calibration, which are not effectively addressed by existing technologies.
Innovation Solution
A system and method that involves down-sampling, equalizing, and up-sampling image volumes to generate a corrected image volume, reducing or removing streak artifacts by determining an error image volume and correcting the original image volume based on it, using techniques like median filtering and Gaussian blurring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiplanar reconstruction (MPR) is performed on CT data, then cross-sectional imaging is achieved, but linear streak artifacts are introduced
Solution Approach 1:
The patent segments the streak artifact removal process into multiple independent steps: detecting streak artifacts, down-sampling the image volume along the streak direction, equalizing the down-sampled volume, up-sampling to restore original resolution, and correcting the original image volume using the error image. This segmentation allows each step to be optimized independently while collectively addressing the streak artifact problem.
Solution Approach 2:
The patent introduces an intermediary error image volume that captures the differences between the original and corrected image volumes. This error image serves as a mediator to guide the correction process, allowing the system to selectively remove streak artifacts while preserving genuine anatomical features through comparative analysis.
2Object-generated harmful factors
If streak artifacts are removed using conventional methods, then image quality improves slightly, but in-plane features are degraded
Solution Approach 1:
The patent applies local quality by treating different spatial regions differently - specifically, applying streak removal operations primarily along the z-direction (where streaks occur) while preserving in-plane (x-y direction) features. The down-sampling and equalization operations are directed along the streak direction, leaving transverse anatomical structures intact.
Solution Approach 2:
The patent transitions from a 2D image processing approach to a 3D volume-based approach. By operating on the third dimension (z-direction) separately through down-sampling and equalization, the system can remove streak artifacts without affecting the two-dimensional in-plane features, effectively adding a dimensional perspective to the correction process.
3Object-generated harmful factors
If image processing operations are applied to reduce streaks, then streak artifacts are reduced, but processing time increases
Solution Approach 1:
The patent performs preliminary down-sampling of the image volume along the z-direction before applying the correction algorithm. This preliminary action reduces the number of pixels that require processing in subsequent steps, thereby decreasing the computational burden and processing time while still maintaining sufficient information for accurate streak removal.
Solution Approach 2:
The patent applies partial action by selectively processing only the z-direction dimension for streak removal, rather than applying uniform processing to all three dimensions. This selective approach reduces the overall processing volume and time while effectively targeting the streak artifact problem in the z-direction.
Data Source
AI summary
A system and method for reducing streak artifacts in a multiplanar reconstruction image are provided. The method may include: retrieving a first image volume from image data, the first image volume including multiple images, at least one of which includes a streak artifact including multiple streaks of a streak width along a first direction; determining a second image volume and a third image volume based on the first image volume; determining an initial error image volume based on the second image volume and the third image volume; determining a revised error image volume based on the initial error image volume; smoothing the revised error image volume to generate a final error image volume; correcting the first image volume according to the final error image volume; and, generating, based on the corrected first image volume, a corrected image volume.