Ultra-limited-angle CT image reconstruction algorithm based on reweighting and edge-preserving.

Area of Science:

• Medical Imaging
• Computational Imaging
• Image Reconstruction

Background:

• Ultra-limited-angle computed tomography (CT) image reconstruction (scanning range ≤ π/2) is severely ill-posed.
• Traditional iterative algorithms struggle with large condition numbers, hindering valid image generation.
• Significant challenges exist in reconstructing high-quality images from limited angular data.

Purpose of the Study:

• To develop and validate a robust algorithm for ultra-limited-angle CT image reconstruction.
• To address the ill-posed nature of reconstruction with restricted angular data.
• To improve the accuracy and quality of CT images obtained under challenging scanning conditions.

Main Methods:

• A novel optimized reconstruction model incorporating a reweighted method to improve the condition number was developed.
• The Reweighted Alternating Edge-preserving Diffusion and Smoothing (AEDS) algorithm was proposed, building upon existing AEDS techniques.
• The proposed algorithm was tested against Pre-Landweber and standard AEDS methods using simulated (Shepp-Logan phantom) and real (flat object) projection data, evaluating with PSNR and SSIM.

Main Results:

• The proposed algorithm demonstrated substantial improvements in image quality for simulated data, increasing PSNR from 22.46 dB to 39.38 dB and SSIM from 0.71 to 0.96.
• For real-world data, the algorithm achieved the highest PSNR (30.89 dB) and SSIM (0.88), yielding a valid reconstructed image.
• Quantitative metrics confirm the superior performance of the new algorithm in ultra-limited-angle CT reconstruction.

Conclusions:

• The developed algorithm effectively combines strengths from various image processing and reconstruction techniques.
• The novel approach significantly outperforms existing methods, proving its validity for ultra-limited-angle CT image reconstruction.
• This work offers a promising solution for generating high-fidelity CT images in scenarios with severe angular data limitations.