Direct Parametric Image Reconstruction of Rapid Multi-tracer PET

Xiaoyin Cheng¹, Nassir Navab², Sibylle I. Ziegler¹, and Kuangyu Shi¹

¹Department of Nuclear Medicine, Technische Universtität München, Germany

²Computer Aided Medical Procedures, Technische Universtität München, Germany

Abstract. The separation of multiple PET tracers within an overlapped scan based on intrinsic difference of pharmacokinetics is challenging due to the limited SNR of PET measurements and high complexity of fitting models. This study developed a novel direct parametric reconstruction method by integrating a multi-tracer model with reduced number of fitting parameters into image reconstruction. To incorporate the multi-tracer model, we adopted EM surrogate functions for the optimization of the penalized log-likelihood. The algorithm was validated on realistic simulation phantoms and real rapid [¹⁸F]FDG and [¹⁸F]FLT PET imaging of mice with lymphoma mouse tumor. Both results have been compared with conventional methods and demonstrated evident improvements for the separation of multiple tracers.

LNCS 8151, p. 155 ff.

Full article in PDF | BibTeX

lncs@springer.com
© Springer-Verlag Berlin Heidelberg 2013