Efficient SPECT scatter calculation in non-uniform media using correlated Monte Carlo simulation

Area of Science:

• Medical Imaging Physics
• Computational Imaging
• Nuclear Medicine Technology

Background:

• Accurate scatter simulation in single photon emission computed tomography (SPECT) is computationally intensive, especially for non-uniform objects.
• Existing methods often require significant computation time and memory, limiting clinical applicability.

Purpose of the Study:

• To develop a computationally efficient method for accurate scatter simulation in SPECT projection data.
• To reduce the computation time and memory requirements for scatter correction in non-uniform media.

Main Methods:

• A novel transform method using slab-derived scatter estimation (P(SDSE)) and Monte Carlo (MC) simulations for uniform (P(u)) and non-uniform (P(nu)) objects.
• The scatter projection P is estimated using the formula P = P(SDSE) * P(nu) / P(u).
• Photon path tracking for P(nu) identical to P(u) and analytical collimator modeling were used to reduce noise.

Main Results:

• The method significantly decreases computation time to a few tens of seconds per projection on a PC.
• Excellent agreement was achieved between the proposed method's scatter projections and standard MC simulations for 99mTc and 201Tl.
• The approach avoids the excessive memory demands of previous 3D model-based scatter correction methods.

Conclusions:

• The developed method offers a computationally efficient and accurate solution for scatter correction in clinical SPECT.
• This technique is attractive for real-time or near-real-time scatter correction, improving image quality and diagnostic accuracy.
• It overcomes the limitations of previous methods regarding computational resources.