Improved FDG kinetic analysis in brain tumors through simultaneous MR/PET acquisition

504 Objectives The goal of the study was to investigate the use of high-resolution, MR-derived arterial input functions (AIF) for improving tracer kinetic parameter estimation on human brain tumors during dynamic FDG injections without direct arterial sampling. Methods PET list mode data were acquired during dynamic injection of FDG in brain tumor patients. Simultaneous acquisition of MR arterial spin labeling (ASL) was also performed during the FDG injection. The PET list mode acquisition was continued for the duration of the entire imaging protocol. Additional MR acquisitions were also performed, including a radial, compressed-sensing-aided, dynamic contrast-enhanced (DCE) contrast study (1). The list mode PET data were reconstructed as 5 second bins for the initial ten minutes and the corresponding PET AIF measured from a region of interest (ROI) in the carotid artery. The time course of the internal carotid artery signal intensity in the DCE scan was used to derive an AIF and this AIF was normalized and scaled to approximate the dispersion of the activity in the blood pool during the dynamic FDG injection (Figure 1). The Sokoloff model (2) was then used to compare the tracer kinetic constants obtained using the MR and PET-based AIF9s. Results As reported elsewhere (3), non-linear fit of the dynamic PET data using PET-based AIF9s was severely biased (table 1) due to the excessive partial voluming present in the images. Using the high-resolution MRI scans to measure the AIF led to a significant decrease in the bias of the estimated tracer kinetic parameters. Using low resolution DCE images also led to increased errors for the calculated parameters, which indicates that partial voluming is an important source of bias. Conclusions Our results support the use of concurrent dynamic MR/PET acquisitions as a means for improved tracer kinetic parameter estimation during FDG clinical examinations in brain tumor patients. Research Support Supported in part by PHS grant R01 CA111996-06.