Smoothed l₀ Norm Regularization for Sparse-View X-Ray CT Reconstruction

Area of Science:

• Medical Imaging
• Image Reconstruction
• Computational Imaging

Background:

• Low-dose computed tomography (CT) reconstruction is crucial for minimizing radiation exposure in medical diagnostics.
• Standard filtered back-projection (FBP) methods often struggle with noise and artifacts at low radiation doses.
• Sparse regularization techniques are increasingly explored to enhance image quality and reduce radiation dose.

Purpose of the Study:

• To develop and evaluate an iterative CT reconstruction method using an improved smoothed L0 (SL0) norm regularization.
• To leverage the sparsity of image gradients for enhanced reconstruction quality.
• To compare the performance of the proposed SL0 method against FBP and total variation (TV) regularization.

Main Methods:

• An iterative reconstruction algorithm incorporating a smoothed L0 (SL0) norm was developed.
• The SL0 norm approximates the L0 norm using continuous functions to exploit image gradient sparsity.
• The method was tested on simulated data (Shepp-Logan phantom, clinical head slice) and real animal experimental data.

Main Results:

• The SL0 regularization method demonstrated superior performance compared to FBP and TV regularization.
• Reconstructed images showed significant noise reduction.
• Crucial anatomical features and tissue details were well-preserved.

Conclusions:

• The proposed iterative reconstruction approach with improved SL0 norm regularization is effective for low-dose CT.
• This method offers a promising solution for improving image quality while reducing radiation dose in CT scans.
• SL0 regularization enhances noise suppression and anatomical feature preservation in medical imaging.