Spatiotemporal reconstruction of list-mode PET data

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Biophysics

Background:

• List-mode positron emission tomography (PET) data offers rich temporal information.
• Accurate estimation of tracer kinetics is crucial for quantitative PET imaging.
• Existing methods may not fully leverage the continuous-time information in list-mode data.

Purpose of the Study:

• To develop and validate a method for continuous time tracer density estimation from list-mode PET data.
• To model count rate functions using cubic B-splines for improved temporal resolution.
• To incorporate spatial and temporal smoothness penalties for robust parameter estimation.

Main Methods:

• Modeling voxel count rates as inhomogeneous Poisson processes with cubic B-spline bases.
• Estimating spline control vertices by maximizing likelihood of photon pair arrival times.
• Applying quadratic spatial and temporal smoothness penalties and nonnegativity constraints.
• Estimating randoms and scatter rate functions using spatiotemporal independence assumptions.

Main Results:

• The method provides a continuous time estimate of tracer density.
• Quantitative evaluation using simulated data demonstrated the method's effectiveness.
• Successful application in a human study using 11C-raclopride.

Conclusions:

• The proposed B-spline based method enables accurate continuous time tracer density estimation from list-mode PET data.
• Incorporation of smoothness penalties enhances the robustness of rate function estimation.
• This approach holds promise for improved kinetic modeling in PET studies.