Application of discrete data consistency conditions for selecting regularization parameters in PET attenuation map...

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Reconstruction

Background:

• Simultaneous emission and transmission (E/T) measurements in PET offer geometric matching and reduced scan times.
• Low transmission statistics can lead to noisy attenuation correction maps, impacting image quality.
• Current regularization parameter selection for noise control is often empirical.

Purpose of the Study:

• To investigate the use of discrete data consistency conditions (DDCC) for optimal selection of regularization parameters in PET attenuation correction.
• To develop an attenuation map reconstruction method that is consistent with emission data and accounts for reconstruction algorithm and acquisition geometry.

Main Methods:

• Utilized discrete data consistency conditions (DDCC) for regularization parameter selection.
• Employed Maximum A Posteriori with Total Variation Regularization (MAP-TR) for attenuation map reconstruction.
• Used 3D Ordered Subset Expectation Maximization (OS-EM) for emission image estimation.
• Validated methodology using a computer-generated whole-body phantom, simulating emission and transmission data.

Main Results:

• DDCC-regularized attenuation maps reduce noise propagation from transmission scans to emission images.
• DDCC prevents over-smoothing of attenuation maps, avoiding resolution mismatch artifacts between emission and transmission data.
• The optimal regularization parameter selection is dependent on emission image resolution, influenced by OS-EM iterations.

Conclusions:

• DDCC provides an optimal and data-driven approach to regularization parameter selection for PET attenuation correction.
• This method enhances the accuracy and reliability of attenuation maps, leading to improved PET image quality.
• DDCC ensures consistency between emission and transmission data, crucial for quantitative PET imaging.