A Model of Regularization Parameter Determination in Low-Dose X-Ray CT Reconstruction Based on Dictionary Learning

Area of Science:

• Medical Imaging
• Radiology
• Computational Imaging

Background:

• X-ray computed tomography (CT) is crucial for anatomical imaging but poses radiation risks.
• Minimizing radiation dose while preserving image quality in CT is a significant challenge.
• Existing low-dose CT methods like total variation (TV) minimization and dictionary learning have limitations, particularly in parameter selection.

Purpose of the Study:

• To develop an improved method for low-dose CT reconstruction.
• To address the challenge of determining critical parameters, such as the regularization parameter, in CT image reconstruction.
• To enhance image quality and reduce radiation exposure in CT scans.

Main Methods:

• Proposed a novel reweighted objective function for CT reconstruction.
• Developed a numerical calculation model for determining the regularization parameter.
• Utilized compressed sensing principles with total variation (TV) minimization and dictionary learning concepts.

Main Results:

• The proposed reweighted objective function demonstrated effective regularization parameter calculation.
• Experimental results showed improved reconstruction image quality compared to existing methods.
• The strategy significantly reduced the time required for CT image reconstruction.

Conclusions:

• The novel reweighted objective function offers a robust solution for low-dose CT reconstruction.
• This approach successfully balances radiation dose reduction with high image quality.
• The method provides a computationally efficient and effective strategy for clinical CT applications.