PT  - JOURNAL ARTICLE
AU  - Si Chen
AU  - Benjamin Tsui
TI  - Evaluation of a new 4D PET image reconstruction method with respiratory motion compensation in a CHO study
DP  - 2011 May 01
TA  - Journal of Nuclear Medicine
PG  - 150--150
VI  - 52
IP  - supplement 1
4099  - http://jnm.snmjournals.org/content/52/supplement_1/150.short
4100  - http://jnm.snmjournals.org/content/52/supplement_1/150.full
SO  - J Nucl Med2011 May 01; 52
AB  - 150 Objectives The objective is to evaluate a new 4D PET image reconstruction method with respiratory motion compensation for the lesion detection tasks using realistic simulation data and the ROC methodology with channelized hotelling observer (CHO). Methods The 4D XCAT phantom which realistically models the anatomies and respiratory motion of a typical patient was used in this study. Six spherical hot lesions of 10mm diameter were inserted into the lungs and liver. Three different FDG distributions were simulated to model variations in patients. The generated 4D phantoms were then input into a Monte-Carlo software combining SimSET and GATE to simulate respiratory-gated PET acquisitions with a Discover LS scanner. Ten noise realizations were simulated for each FDG distribution with or without lesions. The simulated PET datasets were reconstructed by a conventional 3D ML-EM image reconstruction method and a new 4D image reconstruction method with respiratory motion compensation. In the new 4D reconstruction method, the respiratory motion was first estimated based on the respiratory-gated PET sinograms with modeling of the respiration-induced deformation of the attenuation map derived from the single-phase CT image. The estimated respiratory motion was modeled in a 4D OS-EM algorithm to reconstruct a motion-compensated PET image from the respiratory-gated sinograms. All images were then post-filtered with a 3D Gaussian filter. The CHO and ROC methodology were then applied to the post-filtered images to compare the performances of the two reconstruction methods. Results The results showed that, for either lung or liver lesion detection, the area under the ROC curves (AUC) of the new 4D PET image reconstruction method was significantly larger than that of the 3D reconstruction method, whose performance (96% sensitivity at 80% specificity) for lung lesion detection was consistent with clinical studies. Conclusions The new 4D PET image reconstruction method significantly improved the PET image quality by compensating for the patient’s respiratory motion. Research Support NIH grant R01 EB 00016