Dual Domain Optimization Algorithm for CBCT Ring Artifact Correction.

Area of Science:

• Medical Imaging
• Image Processing
• Detector Technology

Background:

• Photon-counting detector (PCD) CT enhances image contrast and reduces radiation dose compared to traditional CT.
• Immature PCD technology leads to stripe artifacts in scanned data, causing ring artifacts in reconstructed CT images.
• Ring artifacts compromise CT value accuracy and introduce pseudo-structures, limiting clinical utility.

Purpose of the Study:

• To develop a novel method for eliminating ring artifacts in PCD-based cone beam CT (CBCT).
• To leverage the unique properties of stripe artifacts in projection data for artifact reduction.

Main Methods:

• Proposed a dual-domain optimization model incorporating group sparsity and low-rank properties of stripe artifacts.
• Developed a TLT (TV-l2,1-Tucker) model specifically for ring artifact removal in PCD-CBCT.
• Designed an efficient iterative algorithm to solve the proposed optimization model.

Main Results:

• Demonstrated that stripe artifacts exhibit group sparsity and low-rank characteristics in 3D projection data.
• The proposed TLT model effectively eliminated ring artifacts in both simulated and real PCD-CBCT data.
• The developed algorithm efficiently solved the optimization problem.

Conclusions:

• The proposed dual-domain optimization model and TLT method significantly outperform existing approaches for ring artifact removal.
• This work enhances the diagnostic accuracy and clinical applicability of PCD-based CBCT imaging.
• The findings pave the way for more reliable and accurate PCD CT imaging.