Tensor Gradient L₀-Norm Minimization-Based Low-Dose CT and Its Application to COVID-19

Area of Science:

• Medical Imaging
• Computational Imaging
• Radiology

Background:

• Low-dose computed tomography (CT) imaging is crucial for reducing radiation exposure.
• Sparse-data subsampling is a key strategy for dose reduction in CT.
• Iterative algorithms are preferred over analytic methods for reconstructing images from sparse CT data due to fewer artifacts.

Purpose of the Study:

• To develop and evaluate a novel Tensor Gradient L0-norm Minimization (TGLM) method for low-dose CT imaging.
• To apply the TGLM method for reconstructing high-quality CT images in Coronavirus Disease 2019 (COVID-19) patients.
• To validate the TGLM method's performance using clinical data from COVID-19 patients.

Main Methods:

• Development of a Tensor Gradient L0-norm Minimization (TGLM) model for low-dose CT.
• Optimization of the TGLM model using the split-Bregman method.
• Application of the TGLM method to low-dose CT scans of COVID-19 patients, incorporating 3-D spatial information.

Main Results:

• The proposed TGLM method successfully reconstructed high-quality CT images from sparse-data subsampled projections.
• The TGLM method demonstrated effectiveness in achieving low-dose scans for COVID-19 detection and severity assessment.
• Validation using clinical data from two COVID-19 patients confirmed the TGLM method's performance.

Conclusions:

• The TGLM method offers a promising approach for high-quality image reconstruction in low-dose CT imaging.
• This technique is particularly beneficial for medical applications like COVID-19 assessment, enabling reduced radiation exposure.
• The TGLM method, optimized with split-Bregman, provides a robust solution for sparse-data CT reconstruction.