Improved iterative reconstruction method for Compton imaging using median filter

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Reconstruction

Background:

• Compton cameras image radio-source distributions without mechanical collimators.
• Ordered-subset expectation-maximization (OS-EM) is standard for Compton image reconstruction but amplifies noise.
• Optimizing OS-EM iterations for high-quality images remains a challenge.

Purpose of the Study:

• To introduce a novel algorithm, Median Root Prior Expectation-Maximization (MRP-EM), to enhance Compton image reconstruction.
• To address the noise amplification and iteration optimization issues inherent in conventional OS-EM algorithms.

Main Methods:

• Implemented a median filter within each iteration of the OS-EM algorithm to create the MRP-EM algorithm.
• Evaluated image quality using mathematical phantoms, assessing spatial resolution, reproducibility, semi-quantitative performance, and uniformity.

Main Results:

• The MRP-EM algorithm effectively reduces noise in reconstructed Compton images.
• MRP-EM demonstrates robustness regarding the number of iterations required for optimal image quality.
• Statistical indices confirm superior performance of MRP-EM compared to conventional reconstruction techniques.

Conclusions:

• The proposed MRP-EM algorithm offers improved noise reduction and image quality for Compton imaging.
• MRP-EM provides a more robust and reliable method for reconstructing radio-source distributions.
• This advancement has the potential to enhance diagnostic accuracy in nuclear medicine applications.