Low-complexity reconstruction of low-dose spectral CT via double low-rank tensor factorization with adaptive...

Area of Science:

• Medical Imaging
• Computational Imaging
• Image Reconstruction

Background:

• Spectral computed tomography (CT) offers enhanced diagnostic information by analyzing X-ray absorption across multiple energy ranges.
• Low-dose spectral CT imaging suffers from significant noise due to insufficient photon counts, compromising image quality.
• Existing reconstruction methods struggle to balance noise suppression and computational efficiency.

Purpose of the Study:

• To develop an advanced sparse-view spectral CT reconstruction technique.
• To effectively suppress mixed noise and reduce computational complexity in low-dose spectral CT.
• To improve the overall quality of reconstructed spectral CT images.

Main Methods:

• Proposed a novel sparse-view spectral CT reconstruction method utilizing a double low-rank tensor factorization framework.
• Incorporated spatial factor denoising, leveraging global spectral low-rankness, group sparsity, and nonlocal self-similarity.
• Employed weighted tensor nuclear norm and weighted group sparsity constraints to exploit image priors.
• Utilized a proximal alternating minimization algorithm for model optimization with convergence guarantees.

Main Results:

• The proposed method effectively suppresses mixed noise in spectral CT images.
• Demonstrated significant reduction in computational complexity compared to existing algorithms.
• Achieved superior spectral CT image quality in both numerical simulations and clinical data.
• The method successfully integrates global, nonlocal, and local image priors.

Conclusions:

• The developed sparse-view spectral CT reconstruction method offers substantial improvements in image quality and computational efficiency.
• This technique shows significant promise for low-dose spectral CT applications, enhancing diagnostic accuracy.
• The integration of tensor factorization and advanced sparsity constraints provides a robust framework for spectral CT reconstruction.