Noise and edge artifacts in maximum-likelihood reconstructions for emission tomography

Area of Science:

• Medical Imaging
• Image Reconstruction
• Tomography

Background:

• Emission tomography image reconstruction often uses maximum-likelihood estimation (MLE).
• The expectation-maximization (EM) algorithm is a common numerical method for achieving MLE.
• As iterations increase, MLE-based images can exhibit noise and edge artifacts.

Purpose of the Study:

• To identify the fundamental cause of noise and edge artifacts in emission tomography reconstructions.
• To develop a novel mathematical approach for artifact suppression in emission tomography.

Main Methods:

• Modification of the standard maximum-likelihood estimation approach.
• Inclusion of specific constraints into the image reconstruction process.
• Mathematical formulation for artifact reduction.

Main Results:

• Demonstrated that artifacts are inherent to the MLE principle, not solely the EM algorithm.
• Developed a constrained mathematical framework to mitigate noise and edge distortions.
• Showcased a method to improve image quality in emission tomography.

Conclusions:

• The observed artifacts in emission tomography are linked to the core maximum-likelihood estimation principle.
• A modified MLE approach incorporating constraints effectively suppresses image noise and edge distortions.
• This constrained method offers improved image fidelity for emission tomography applications.