Fully three-dimensional OSEM-based image reconstruction for Compton imaging using optimized ordering schemes.

Area of Science:

• Medical Imaging
• Nuclear Physics
• Computational Science

Background:

• The ordered subset expectation maximization (OSEM) algorithm is widely used for image reconstruction in Compton cameras.
• OSEM's performance depends on balanced subsets, which is challenging with Compton scattered data.
• Existing methods for subset balancing may not be optimal for Compton camera applications.

Purpose of the Study:

• To compare different data ordering schemes for balancing subsets in OSEM reconstruction for Compton cameras.
• To evaluate the effectiveness of scattering angles (SA), detected positions (DP), and their combination (AP) for data allocation.
• To identify the optimal ordering scheme for accurate Compton camera imaging.

Main Methods:

• Investigated three data ordering schemes: random ordering scheme (ROS), multilevel ordering scheme (MLS), and weighted-distance ordering scheme (WDS).
• Allocated Compton scattered data into J subsets using SA, DP, or AP.
• Calculated coefficients of variation (CVs) and percentage errors to assess subset balance and image accuracy.

Main Results:

• All ordering schemes accelerated reconstruction compared to standard EM with similar computation times.
• SA-based MLS and DP-based WDS yielded the most balanced subsets for their respective arrangements.
• The combination of AP and WDS demonstrated the lowest percentage errors, indicating superior image reconstruction.

Conclusions:

• The weighted-distance ordering scheme (WDS) is effective in creating balanced subsets for Compton camera OSEM reconstruction.
• Combining scattering angle and detected position data (AP) with WDS provides the most accurate reconstructions.
• The AP-WDS combination is likely the optimal strategy for Compton camera image reconstruction using OSEM.