Sparsity-induced dynamic guided filtering approach for sparse-view data toward low-dose x-ray computed tomography.

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computed Tomography

Background:

• Iterative reconstruction (IR) methods are crucial for computed tomography (CT).
• Total variation (TV) regularization effectively reduces artifacts and noise in sparse-view CT.
• Limited-view CT scanning presents challenges, including blurred edges and blocky artifacts, which TV regularization struggles to suppress.

Purpose of the Study:

• To introduce a novel method, sparsity-induced dynamic guided image filtering reconstruction (SIDGIFR), to enhance image quality in limited-view CT.
• To address the limitations of existing methods in preserving edge structures and suppressing artifacts.

Main Methods:

• The proposed SIDGIFR method utilizes intermediate reconstruction results from total difference (TD) minimization as a guidance image.
• Guided image filtering (GIF) is employed to filter projection onto convex sets (POCS) results using the dynamically updated guidance image.
• The dynamic updating of the guidance image facilitates the transfer of crucial features like edges and details throughout the iterative process.

Main Results:

• Simulated and real data studies demonstrate the efficiency and feasibility of the SIDGIFR algorithm.
• Quantitative evaluations show superior performance of SIDGIFR compared to classical IR methods.
• The SIDGIFR algorithm excels at preserving edge structures while effectively suppressing noise and artifacts.

Conclusions:

• The SIDGIFR method offers significant improvements in image quality for limited-view CT reconstruction.
• This technique provides better edge preservation and artifact suppression than existing IR methods.
• SIDGIFR represents a promising advancement for high-quality CT imaging with limited projection data.