Tomosynthesis Image Processing Reducing Ripple Artifacts via Frequency Decomposition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Tomosynthesis imaging is limited by artifacts such as ripple artifacts, which degrade the quality of pseudo two-dimensional images generated from tomographic images, and existing techniques do not adequately improve the quality of these images.
Innovation Solution
An image processing apparatus that decomposes projection images into low-frequency and high-frequency components, applies different image processing techniques to each component, and generates pseudo two-dimensional images using reconstructed tomographic images to reduce artifacts and noise, thereby improving image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If tomosynthesis imaging is performed with limited irradiation angles, then the imaging process is simple and quick, but ripple artifacts and residual images are generated in the reconstructed tomographic images
Solution Approach 1:
The patent segments the projection images into low-frequency components and high-frequency components using spatial frequency separation. Different image processing methods are applied to each component: artifact reduction processing is applied to low-frequency components to suppress ripple artifacts, while noise reduction processing is applied to high-frequency components. This segmentation allows simultaneous optimization of image quality and processing efficiency.
Solution Approach 2:
The patent applies different image processing methods to different frequency components of the projection images. Low-frequency components undergo artifact reduction processing to eliminate ripple artifacts, while high-frequency components undergo noise reduction processing to preserve fine details. This local quality approach ensures that each frequency component is processed optimally, resolving the contradiction between imaging speed and image quality.
2Loss of information
If multiple projection images are superimposed to reconstruct tomographic images, then three-dimensional information is obtained, but artifacts such as ripple artifacts indicating virtual images are generated
Solution Approach 1:
The patent segments projection images into low-frequency and high-frequency components, applying artifact reduction processing specifically to low-frequency components where ripple artifacts predominantly occur. This segmentation enables targeted artifact suppression while preserving the three-dimensional information contained in the multi-angle projection images.
Solution Approach 2:
The patent converts the harmful ripple artifacts into a manageable problem by separating them into the low-frequency component through spatial frequency decomposition. The artifact reduction processing then specifically targets and eliminates these artifacts from the low-frequency component, while the high-frequency component preserves the fine details and structural information, effectively converting the artifact problem into a solvable frequency separation task.
3Manufacturing precision
If artifact reduction processing is applied to reduce ripple artifacts, then image quality improves, but noise may be amplified
Solution Approach 1:
The patent segments the projection images into low-frequency and high-frequency components, applying artifact reduction processing only to the low-frequency component where ripple artifacts are present. The high-frequency component undergoes noise reduction processing instead, preventing noise amplification while preserving fine details. This frequency-based segmentation resolves the contradiction between artifact reduction and noise control.
Solution Approach 2:
The patent applies different processing strategies to different frequency components: artifact reduction is applied locally to low-frequency components where ripple artifacts occur, while noise reduction is applied locally to high-frequency components where fine details and noise are present. This local quality approach ensures that artifact reduction does not amplify noise, as each frequency band receives appropriately tailored processing.
4Device complexity
If a single image processing method is applied to all projection images, then the processing is simple and fast, but the quality of the generated pseudo two-dimensional image is degraded
Solution Approach 1:
The patent segments projection images into low-frequency and high-frequency components, applying different processing methods to each component. Low-frequency components receive artifact reduction processing while high-frequency components receive noise reduction processing. This segmentation enables optimized image quality without excessive complexity, as the processing is automated and based on objective frequency criteria.
Solution Approach 2:
The patent changes the processing parameters based on the spatial frequency content of each projection image component. By automatically identifying and separating low-frequency and high-frequency components, the system dynamically adjusts processing parameters to apply appropriate artifact reduction or noise reduction techniques, achieving high image quality without manual intervention or excessive complexity.
Data Source
AI summary
A control unit of a console acquires a plurality of projection images obtained by irradiating a subject disposed between a radiation source and a radiation detector with radiation emitted from the radiation source at different irradiation angles and capturing the radiation with the radiation detector at each of the irradiation angles and decomposes each of the plurality of projection images into a plurality of first projection images with a low-frequency component lower than a predetermined spatial frequency and a plurality of second projection images with a high-frequency component higher than the predetermined spatial frequency. The control unit generates a pseudo two-dimensional image using a plurality of first tomographic images and a plurality of second tomographic images.