X-ray CT metal artifact reduction using wavelet domain L0 sparse regularization.

Area of Science:

• Medical Imaging
• Image Processing
• Computational Science

Background:

• X-ray computed tomography (CT) imaging is crucial for medical diagnosis.
• Metallic implants in patients often cause severe streaking artifacts in CT scans, hindering accurate diagnosis.
• Existing metal artifact reduction (MAR) methods have limitations in effectively removing these artifacts.

Purpose of the Study:

• To develop and evaluate a new projection completion MAR algorithm for CT imaging.
• To address the challenge of streaking artifacts caused by metallic implants.
• To improve the diagnostic quality of CT images in the presence of metal hardware.

Main Methods:

• A novel MAR algorithm based on projection completion in the wavelet domain was proposed.
• The Douglas-Rachford splitting (DRS) algorithm was employed to solve the regularized inverse problem.
• The algorithm utilized prior information, including wavelet coefficient sparsity and prior sinogram details, with a pseudo-L0 synthesis prior.

Main Results:

• The proposed L0-DRS MAR algorithm demonstrated substantial suppression of streaking artifacts.
• Both simulated and clinical studies (hip prostheses, dental fillings, spine fixation, brain electrodes) were used for evaluation.
• The algorithm outperformed standard linear interpolation and the normalized metal artifact reduction (NMAR) approach.

Conclusions:

• The developed L0-DRS MAR algorithm is effective in reducing metal artifacts in X-ray CT.
• This method offers a significant improvement over existing MAR techniques.
• The algorithm holds promise for enhancing the quality of medical CT imaging with metallic implants.