Robust Low-Dose CT Perfusion Deconvolution via Tensor Total-Variation Regularization.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Radiology

Background:

• Acute brain diseases like stroke are leading causes of death, making rapid diagnosis critical.
• CT Perfusion (CTP) is vital for treatment but high radiation doses and low-dose noise pose challenges.
• Accurate hemodynamic parameter estimation is crucial for timely thrombolytic therapy.

Purpose of the Study:

• To develop a robust and efficient framework for accurate perfusion parameter estimation at low radiation doses.
• To address the limitations of current CTP methods concerning radiation exposure and image quality.
• To improve patient safety and diagnostic accuracy in acute cerebrovascular disease management.

Main Methods:

• Development of a tensor total-variation (TTV) technique integrating spatial and temporal data.
• Proposal of an efficient algorithm for fast convergence and reduced computational complexity.
• Evaluation using digital phantoms and in vivo clinical subjects to assess noise sensitivity, accuracy, and contrast preservation.

Main Results:

• Reduced radiation dose to 8% of the original level.
• Outperformed state-of-the-art algorithms with a 32% improvement in peak signal-to-noise ratio.
• Corrected over-estimation of cerebral blood flow (CBF) and under-estimation of mean transit time (MTT), reducing oscillations.

Conclusions:

• The TTV framework provides accurate and robust perfusion parameter estimation under low-dose CTP conditions.
• This method enhances patient safety by significantly reducing radiation exposure.
• The framework improves diagnostic capabilities for acute brain diseases, maintaining distinction between affected and normal brain regions.