On the relation between the ISRA and the EM algorithm for positron emission tomography.

Area of Science:

• Medical Imaging
• Computerized Tomography
• Image Reconstruction Algorithms

Background:

• The expectation maximization (EM) algorithm is a standard for image reconstruction in emission computed tomography.
• Physical considerations are crucial for refining reconstruction algorithms in medical imaging.

Purpose of the Study:

• To introduce the image space reconstruction algorithm (ISRA) as a modification of the EM algorithm.
• To adapt ISRA for penalized likelihood estimates in emission computed tomography.

Main Methods:

• Developed ISRA based on physical principles, modifying the EM algorithm.
• Established a common mathematical framework for both ISRA and EM algorithms.
• Extended ISRA to incorporate penalized likelihood estimation.

Main Results:

• ISRA converges to least squares solutions, unlike the EM algorithm's focus on Poisson likelihoods.
• Demonstrated that ISRA and EM algorithms share a unifying mathematical foundation.
• Successfully extended ISRA for penalized likelihood estimates, enhancing its applicability.

Conclusions:

• ISRA provides a valuable alternative to the EM algorithm for emission computed tomography reconstruction.
• The shared mathematical framework facilitates algorithm development and extension.
• Penalized likelihood estimation in ISRA offers improved image reconstruction capabilities.