A generalized Gibbs prior for maximum a posteriori reconstruction in SPECT

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Reconstruction

Background:

• Maximum a posteriori (MAP) reconstruction is crucial for improving image quality in single-photon emission computed tomography (SPECT).
• Existing Gibbs priors offer limited flexibility in controlling image characteristics during reconstruction.
• Selecting appropriate prior parameters remains a challenge in clinical SPECT applications.

Purpose of the Study:

• To introduce a novel three-parameter Gibbs prior for enhanced MAP reconstruction in SPECT.
• To demonstrate the flexibility of the new prior in approximating existing methods and offering a wider range of image property emphasis.
• To investigate the impact of the Gibbs prior's derivative potential function (DPF) on image estimation.

Main Methods:

• Development of a new three-parameter Gibbs prior for SPECT image reconstruction.
• Comparative analysis of the new prior against existing two-parameter priors.
• Demonstration of parameter effects on reconstructed SPECT images using acquired data.
• Analysis of the derivative potential function (DPF) in influencing image properties.

Main Results:

• The three-parameter Gibbs prior successfully approximates results from two-parameter priors and a continuum of others.
• Increased user flexibility is achieved in emphasizing desired properties (e.g., edge sharpness, smoothing, low-contrast detail) in reconstructed SPECT images.
• The derivative potential function (DPF) significantly influences the emphasized properties in iterated image estimates.

Conclusions:

• The novel three-parameter Gibbs prior offers significant advantages in SPECT image reconstruction due to its flexibility.
• Careful selection of the three parameters is essential, considering object characteristics and desired image features.
• This enhanced prior provides a valuable tool for optimizing SPECT image quality in clinical practice.