DuDoDR-Net: Dual-domain data consistent recurrent network for simultaneous sparse view and metal artifact reduction...

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computational Imaging

Background:

• Sparse-view computed tomography (SVCT) reduces radiation dose but causes artifacts.
• Metallic implants exacerbate artifacts in SVCT, hindering diagnosis.
• Existing methods for SVCT or metal artifact reduction (MAR) are insufficient for combined sparse-view and metal artifact reduction (SVMAR).

Purpose of the Study:

• To develop an effective method for simultaneous sparse-view and metal artifact reduction (SVMAR).
• To address the under-explored challenge of reconstructing high-quality images from limited projections in the presence of metallic artifacts.

Main Methods:

• Proposed DuDoDR-Net, a dual-domain data-consistent recurrent network.
• Employed recurrent restorations in both image and sinogram domains.
• Introduced image data consistent layer (iDCL) and sinogram data consistent layer (sDCL) for preserving projection data.

Main Results:

• DuDoDR-Net successfully reconstructed artifact-free images.
• The method preserved anatomical structures effectively.
• Achieved superior performance compared to existing SVCT and SVMAR techniques across various sparse-view settings.

Conclusions:

• DuDoDR-Net offers a robust solution for SVMAR.
• The proposed network significantly improves image quality in challenging sparse-view CT scenarios with metallic implants.
• This advancement has potential implications for improved medical diagnosis and downstream applications.