X-ray CT Image Reconstruction Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing X-ray CT image reconstruction methods struggle to simultaneously reduce granular noise and streak-like artifacts while maintaining clear structure boundaries, as prior techniques either obscure the structure boundary when prioritizing streak-like artifact reduction or fail to sufficiently reduce streak-like artifacts.
Innovation Solution
An image reconstruction apparatus and method that separately address granular noise and streak-like artifacts by generating a streak reduction image and an addition image, using distinct noise reduction ratios to reduce noise while retaining structure boundaries, and adding the addition image to the streak reduction image to achieve both noise reduction and clear boundaries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If a single noise reduction method is applied to reduce both granular noise and streak-like artifact, then noise reduction is achieved, but structure boundary becomes obscured
Solution Approach 1:
The patent segments the noise reduction process into two distinct pathways: one for granular noise reduction (maintaining structure boundary) and another for streak-like artifact reduction. By dividing the single noise reduction task into separate handling for different noise types, the patent achieves comprehensive noise reduction while preserving structure boundary clarity through the addition image that retains high-frequency components.
Solution Approach 2:
The patent applies different noise reduction ratios locally: a first noise reduction ratio for granular noise that preserves structure boundary, and a second noise reduction ratio for streak-like artifact that can be more aggressive. This local differentiation allows each noise type to be treated with appropriate reduction intensity, resolving the contradiction between overall noise reduction and structure boundary preservation.
2Object-affected harmful factors
If noise reduction ratio is increased to reduce streak-like artifact, then streak-like artifact reduction is improved, but structure boundary becomes obscured
Solution Approach 1:
The patent separates streak-like artifact reduction into a dedicated streak reduction image generation pathway with its own noise reduction ratio (second noise reduction ratio). This segmentation allows aggressive streak reduction without affecting structure boundary, as the boundary information is preserved in the addition image generated separately with a different noise reduction ratio (first noise reduction ratio).
Solution Approach 2:
The addition image acts as an intermediary that carries structure boundary information. By generating the addition image with a first noise reduction ratio that preserves boundaries, and then adding it to the streak reduction image, the patent mediates between aggressive streak reduction and boundary preservation, allowing the latter to benefit from the former's boundary-preserving properties.
3Manufacturing precision
If noise reduction ratio is decreased to retain structure boundary, then structure boundary clarity is improved, but streak-like artifact reduction is insufficient
Solution Approach 1:
The patent segments the noise reduction task into two parallel processes: one generating a streak reduction image with a second noise reduction ratio optimized for artifact removal, and another generating an addition image with a first noise reduction ratio optimized for boundary preservation. This segmentation allows each process to optimize for its specific goal without compromising the other.
Solution Approach 2:
The patent merges the streak reduction image and the addition image through addition operation. The streak reduction image provides aggressive artifact removal, while the addition image contributes preserved structure boundary information. By combining these two images, the patent achieves both sufficient streak-like artifact reduction and clear structure boundary retention simultaneously.
Data Source
AI summary
The present invention provides an image reconstruction apparatus and an image reconstruction method, both of which achieve reduction in a granular noise and a streak-like artifact both contained in a reconstruction image, and a clear structure boundary as well. The image reconstruction apparatus configured to generate a reconstruction image includes a noise reduction section that generates a streak reduction image as an image obtained by reducing a streak-like artifact in the reconstruction image, a noise reduction ratio acquisition section that acquires a first noise reduction ratio as a noise reduction ratio to be used for reducing a granular noise while retaining a structure boundary in the reconstruction image, and a second noise reduction ratio as a noise reduction ratio to be used for generating the streak reduction image, and an image addition section that adds an addition image to be generated based on the first noise reduction ratio and the second noise reduction ratio to the streak reduction image.