Weighted expectation maximization reconstruction algorithms with application to gated megavoltage tomography

Area of Science:

• Medical physics
• Image processing
• Computational imaging

Background:

• X-ray tomography is crucial for medical imaging.
• Conventional expectation maximization (EM) algorithms struggle with limited and unevenly sampled data, leading to artifacts.
• Respiratory-gated megavoltage tomography presents unique challenges due to asymmetric, limited, and unevenly sampled projection data.

Purpose of the Study:

• To develop and investigate weighted expectation maximization (EM) algorithms for enhanced image reconstruction in X-ray tomography.
• To address the specific challenges of respiratory-gated megavoltage tomography, including data inconsistencies and truncation.
• To mitigate ring- and streak-like artifacts common in conventional EM reconstructions.

Main Methods:

• Implementation of weighted expectation maximization (EM) algorithms tailored for X-ray tomography.
• Utilizing computer-simulated and clinical gated fan-beam megavoltage projection data for validation.
• Comparative analysis of weighted EM algorithms against conventional EM methods.

Main Results:

• Weighted EM algorithms effectively reduced ring- and streak-like artifacts in reconstructed images.
• Demonstrated mitigation of image artifacts caused by data inconsistencies and truncation.
• Improved image quality in simulated and clinical respiratory-gated megavoltage tomography datasets.

Conclusions:

• Weighted EM algorithms offer a significant improvement over conventional EM for X-ray tomography image reconstruction.
• The proposed algorithms are particularly effective for challenging datasets, such as those in respiratory-gated megavoltage imaging.
• This work provides a robust solution for enhancing image quality and reducing artifacts in specialized tomographic applications.