Segmental limited-angle CT reconstruction based on image structural prior

Area of Science:

• Medical Imaging
• Computational Imaging
• Image Reconstruction

Background:

• X-ray CT reconstruction from incomplete data, particularly few-view and limited-angle sampling, presents significant challenges.
• Few-view sampling requires rapid tube voltage switching, posing technical demands, while limited-angle sampling often results in image artifacts.
• Existing methods struggle with artifact suppression and maintaining image quality under sparse data conditions.

Purpose of the Study:

• To investigate a novel segmental limited-angle (SLA) sampling strategy for X-ray CT.
• To develop a reconstruction method that incorporates image structural priors to mitigate artifacts.
• To demonstrate the effectiveness of the proposed SLA sampling and reconstruction approach for improved CT image quality.

Main Methods:

• A new segmental limited-angle (SLA) sampling strategy was developed, avoiding rapid tube voltage switching.
• An image reconstruction model incorporating image structural priors was implemented to suppress artifacts.
• The proposed method was tested using digital phantoms and real-world data (carved cheese, walnut).

Main Results:

• The SLA sampling strategy resulted in projection data with lower correlation, beneficial for reconstruction.
• Reconstruction experiments showed the proposed method produced images closer to reference images.
• Quantitative evaluation using RMSE, PSNR, and SSIM demonstrated the superiority of the method in simulation.
• Real CT data reconstructions confirmed effective artifact reduction and preservation of image structures.

Conclusions:

• The proposed segmental limited-angle (SLA) sampling strategy offers an effective alternative to traditional limited-angle CT.
• Incorporating image structural priors into the reconstruction model significantly suppresses artifacts.
• The method successfully reconstructs high-quality CT images from sparse projection data, preserving fine details.