Kinetic modeling of 18F-FMISO in glioblastoma

We reported in this work the modeling of FMISO in human glioblastoma with compartment modeling. Because FMISO behaves differently in hypoxic tumors than in perfused tumors and normal tissues, it appears mandatory to determine hypoxic from perfused tissues. Spectral analysis could be a safe tool to fulfill this requirement. We have observed hypoxia in the images by means of kinetic modeling but the determination of the hypoxia by means of tumor to blood ratio was not always accurate. Kinetic modeling of FMISO could be made with the usual 3-compartment model. Our measurements were made dynamic for 15 min completed with 10 min scans at 2 h, 3 h and 4 h, and in other scans for 30 min dynamic completed with scans at 2 h and 3 h as requested by the clinic. Bruelmeier et al. used 90 min scans. We have shown that scanning for 2 h or 4 h are apparently similar. Reducing the scan duration to 60 or 90 min would be efficient in reducing errors in image misregistration and in increasing detection efficiency. Kinetic modeling with a two-compartment model and with a three-compartment model when forcing k4 to be 0 did not perform good fits to the pixel data.