4D Medical Imaging Reconstruction with Selective Denoising
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing time-resolved medical imaging techniques face challenges in achieving high image quality, particularly in low-contrast regions, while minimizing X-ray dose and acquisition time, as current denoising methods can negatively impact spatial resolution and geometrical accuracy.
Innovation Solution
A method that assigns imaging datasets to two groups based on noise-affecting parameters, denoising one group and using both groups to reconstruct a 4D-volume, optimizing SNR and dose by varying imaging parameters like X-ray dose or acquisition time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If denoising algorithms are applied to improve image quality, then signal-to-noise ratio is improved, but spatial resolution and geometrical accuracy deteriorate
Solution Approach 1:
The imaging datasets are divided into two groups: a first group with lower noise affecting parameters that undergoes denoising, and a second group with higher noise affecting parameters that remains un-denoised. This segmentation allows selective application of denoising to only those datasets where it provides benefit, preserving spatial resolution in the un-denoised group while improving SNR in the denoised group.
Solution Approach 2:
Different quality processing is applied to different groups of imaging datasets based on their specific noise characteristics. The first group receives denoising enhancement while the second group maintains original quality, creating local optimization where each group's specific needs are addressed differently in the final 4D-volume reconstruction.
2Measurement precision
If X-ray dose is increased to improve image quality, then signal-to-noise ratio is improved, but patient radiation exposure worsens
Solution Approach 1:
The noise affecting parameter (such as X-ray dose or acquisition time) is varied across different imaging datasets to create distinct groups. By changing this parameter systematically, the method optimizes the balance between image quality and patient exposure, using statistical reconstruction to combine data from multiple parameter settings.
Solution Approach 2:
The method uses statistical reconstruction that incorporates information from both denoised and un-denoised datasets, effectively using the un-denoised high-dose data to inform and improve the denoised low-dose data, creating a feedback loop that optimizes overall image quality while minimizing total dose.
3Measurement precision
If acquisition time is prolonged to improve image quality, then signal-to-noise ratio is improved, but temporal resolution and patient comfort worsen
Solution Approach 1:
The acquisition process is segmented into multiple shorter scans with different noise affecting parameters rather than one long scan. This allows the total acquisition time to be distributed across multiple brief acquisitions, improving SNR through statistical combination while maintaining temporal resolution and patient comfort.
Data Source
Figure 1~2
Figure 3
AI summary
For generating a reconstructed 4D-volume in time-resolved medical imaging, a plurality of temporally ordered imaging datasets (5a, 5b) representing an imaged object corresponding to a motion of the object is received. Each imaging dataset of the plurality of imaging datasets (5a, 5b) is assigned to a first group (5a) or to a second group (5b) depending on a noise-affecting imaging parameter used for generating the respective imaging dataset. Each imaging dataset of the first group (5a) is denoised using a denoising algorithm. A 4D-volume of the object is generated based on each denoised imaging dataset of the first group (5a) and based on each imaging dataset of the second group (5b).