Sinogram smoothing techniques for myocardial blood flow estimation from dose-reduced dynamic computed tomography.

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Radiology

Background:

• Dynamic contrast-enhanced computed tomography (CT) is valuable for myocardial blood flow (MBF) estimation in coronary artery disease.
• High radiation dose is a significant limitation of current dynamic CT techniques.
• Reducing radiation dose via lower tube current or fewer temporal frames introduces data noise.

Purpose of the Study:

• To investigate data-domain smoothing techniques for improving MBF estimation from low-dose dynamic CT acquisitions.
• To compare the efficacy of sinogram restoration and Karhunen-Loeve (KL) transform smoothing for MBF quantification.

Main Methods:

• Simulated low-dose dynamic CT cardiac acquisitions were used.
• Two smoothing techniques were evaluated: sinogram restoration (spatial and temporal) and KL transform (temporal smoothing in sinogram domain).
• A quantitative two-compartment blood flow model was applied to estimate MBF from time-attenuation curves.

Main Results:

• Sinogram smoothing generally yielded better MBF estimates across various dose levels.
• The KL transform method outperformed sinogram smoothing at the lowest simulated dose levels (25 mAs, 2-3s temporal spacing).
• KL transform provided improved MBF estimates compared to conventional processing only at very low doses (<7 mSv).

Conclusions:

• Proposed smoothing techniques, particularly sinogram restoration and KL transform, can enable substantial radiation dose reduction in dynamic CT for MBF estimation.
• These methods offer potential for high-fidelity MBF information even with reduced radiation exposure.
• The KL transform is particularly effective for MBF estimation in ultra-low dose scenarios.