Backprojection regularization with weighted ramp filter for tomographic reconstruction.

Area of Science:

• Medical Imaging
• Computational Imaging
• Image Reconstruction

Background:

• Filtered backprojection (FBP) is a common CT reconstruction method.
• Backprojection then filtering (BPF) is less common due to reconstruction quality issues.
• BPF suffers from DC shift, aliasing artifacts, and high-frequency noise amplification.

Purpose of the Study:

• To develop an improved backprojection-based CT reconstruction method.
• To address the limitations of traditional BPF, specifically DC shift and noise amplification.
• To enhance low-dose computed tomography (CT) imaging reconstruction quality.

Main Methods:

• Investigated a weighted ramp filter to reduce high-frequency noise amplification.
• Developed a total-variation based backprojection regularization (BPR) method.
• Mitigated DC shift and improved noise robustness in BPF.

Main Results:

• The developed BPR method demonstrated superior performance compared to FBP.
• BPR effectively reduced noise amplification and mitigated DC shift artifacts.
• Experimental results confirmed BPR's advantage in low-dose CT imaging.

Conclusions:

• Backprojection regularization (BPR) offers a significant improvement over FBP for low-dose CT.
• The novel weighted ramp filter and TV-based regularization enhance reconstruction quality.
• BPR provides a more robust and higher-fidelity solution for CT image reconstruction.