Statistical image reconstruction from limited projection data with intensity priors.

Area of Science:

• Medical Imaging
• Radiology
• Image Reconstruction

Background:

• X-ray computed tomography (CT) scans pose a radiation dose risk, increasing malignancy concerns.
• Low-dose CT protocols are sought to minimize patient exposure without compromising image quality.
• Statistical reconstruction (SR) offers superior image quality over analytical methods but faces computational challenges.

Purpose of the Study:

• To develop a novel framework for statistical reconstruction in x-ray CT with low angular sampling.
• To address streak artifacts in low-dose CT by incorporating prior anatomical and attenuation information.
• To improve image quality in low-dose CT reconstruction.

Main Methods:

• Proposed a statistical reconstruction framework for x-ray CT utilizing low angular sampling.
• Incorporated prior knowledge of object structures and attenuation into the image reconstruction objective function.
• Formulated the objective function using the ℓ(1) norm distance, informed by compressed sensing principles.

Main Results:

• The proposed method effectively suppresses streak artifacts in low-dose CT images.
• Experimental studies on simulated and real data demonstrated significant improvements in image quality.
• The framework leverages easily computable prior information from homogeneous regions.

Conclusions:

• The developed framework offers a viable solution for high-quality image reconstruction in low-dose x-ray CT.
• Incorporating prior intensity information is effective in mitigating artifacts caused by low angular sampling.
• This approach enhances the utility of low-dose CT protocols in clinical settings.