Spatial-Temporal PET Reconstruction with CT-Derived Organ Deformation

 

Tianfang Li and Lei Xing

Department of Radiation Ocology

Stanford University, School of Medicine

Stanford, CA 94305-5847

 

Four-dimensional (4D) PET (acquired with gated, dynamic or list mode) has been used for tumor delineation and evaluation of response to radiotherapy for a various cancers. However, the quantitative accuracy of 4D PET is limited due to its poor statistics, since the total coincidence counts in the conventional 3D acquisition are now divided into many phase bins and each of them is treated as an independent entity in 4D imaging. In this work, we develop a mathematically rigorous approach to maximally enhance the signal-to-noise ratio of 4D PET by simultaneously considering the coincidences acquired at all time points when reconstructing the phase-resolved images. By deformable registration of 4D-CT images, a patient-specific motion model was derived and incorporated into our ¡°spatial-temporal PET reconstruction¡± algorithm based on the maximum likelihood principle. Via a novel concept of ¡°virtual curved line-of-response¡±, we show that the PET ¡°4D likelihood¡± can be maximized with a modified expectation-maximization algorithm. The approach was quantitatively evaluated with numerical and physical phantom experiments. Five clinical studies of pancreatic, lung and liver cancer patients were then carried out. From these studies, it is found that the quantitative accuracy can be reached usually within 40 iterations and the SNRs can be increased more than 80% over the regular 4D PET and 35% over 3D PET. The new spatial-temporal reconstruction formalism allow us to fully take advantage of the information acquired with combined PET/CT scanner and obtain a substantially improved 4D PET imaging.