Statistical sinogram restoration in dual-energy CT for PET attenuation correction

Area of Science:

• Medical Imaging Physics
• Radiological Sciences
• Computational Imaging

Background:

• Dual-energy (DE) X-ray computed tomography (CT) offers valuable applications in medical imaging.
• Low-dose DECT is promising for attenuation correction in positron emission tomography (PET) imaging.
• Current sinogram material decomposition methods for DECT are limited by noise and logarithmic transforms.

Purpose of the Study:

• To develop novel sinogram restoration methods for low-dose dual-energy CT (DECT).
• To improve the accuracy of attenuation correction in positron emission tomography (PET) imaging.
• To reduce noise in component sinogram estimates derived from low-dose DECT data.

Main Methods:

• Proposed two novel sinogram restoration techniques: penalized weighted least square (PWLS) and penalized likelihood (PL).
• These methods are based on statistical models to address noise characteristics in DECT data.
• Simulations were conducted to compare the proposed methods against existing classical approaches.

Main Results:

• The proposed PWLS and PL methods yield significantly less noisy component sinogram estimates compared to classical techniques.
• These improved sinogram estimates lead to more precise attenuation correction in PET emission images.
• Simulations demonstrated the superiority of the novel statistical methods for low-dose DECT applications.

Conclusions:

• Novel statistical sinogram restoration methods (PWLS and PL) enhance low-dose DECT material decomposition.
• Improved attenuation correction in PET imaging is achieved, leading to better diagnostic quality.
• The proposed techniques offer a significant advancement over existing methods for DECT-based PET attenuation correction.