Level set method for positron emission tomography

Area of Science:

• Medical Imaging
• Computational Science

Background:

• Positron Emission Tomography (PET) relies on radioactive tracers to visualize biological processes.
• Expectation Maximization (EM) algorithms are standard for PET image reconstruction, estimating tissue concentration coefficients.
• Current methods may struggle with capturing fine details and incorporating anatomical context.

Purpose of the Study:

• To develop an improved PET image reconstruction algorithm.
• To integrate anatomical information seamlessly into the reconstruction process.
• To enhance the estimation of tissue concentration coefficients.

Main Methods:

• A novel algorithm combining the Expectation Maximization (EM) algorithm with a level set approach.
• Utilizing a multiple level set formulation to represent object geometry.
• Incorporating anatomical information naturally within the level set framework.

Main Results:

• The combined EM and level set method effectively captures coarse-scale information and discontinuities in concentration coefficients.
• Anatomical information is efficiently integrated, improving reconstruction accuracy.
• The proposed algorithm demonstrates applicability across various PET configurations.

Conclusions:

• The integration of level set methods with EM algorithms offers a powerful new approach for PET image reconstruction.
• This technique enhances the ability to represent object geometry and incorporate anatomical data.
• The algorithm provides a flexible and effective solution for diverse PET imaging scenarios.