Effects of sparse sampling schemes on image quality in low-dose CT

Area of Science:

• Medical Imaging
• Radiology
• Computational Imaging

Background:

• Low-dose computed tomography (CT) aims to reduce radiation exposure risks.
• Compressive-sensing (CS) algorithms reconstruct images from sparse data, showing promise for low-dose CT.
• The impact of sparse sampling schemes on CS-based image quality remains understudied.

Purpose of the Study:

• To present and analyze sparse-sampling schemes for low-dose CT.
• To quantitatively assess data properties of these schemes.
• To compare the effects of different sampling schemes on image quality in CS reconstruction.

Main Methods:

• Analysis of sampling density and data incoherence for CS-based reconstruction.
• Simulation of five different sparse sampling schemes targeting dose reduction (75% and 87.5%).
• Numerical realization of sparse sampling based on a fully sampled circular cone-beam CT dataset.

Main Results:

• Both sampling density and data incoherence critically influence CS reconstruction image quality.
• Sparse-view, MVUS-fine, and MVUS-moving schemes demonstrated promising results.
• These schemes achieved image quality comparable to the reference, with structure similarity index > 0.92 at 75% dose reduction.

Conclusions:

• Sampling density and data incoherence are key factors affecting image quality in CS reconstruction.
• Optimizing sampling schemes using these indicators is crucial for effective low-dose CT.
• This approach enables the acquisition of optimally sampled sparse data for superior CS algorithm performance.