Hybrid [18F]-FDG PET/MRI including non-Gaussian diffusion-weighted imaging (DWI): Preliminary results in non-small cell lung cancer (NSCLC)

Abstract

PurposeTo assess the feasibility of non-Gaussian DWI as part of a FDG-PET/MRI protocol in patients with histologically proven non-small cell lung cancer. Material and methods15 consecutive patients with histologically proven NSCLC (mean age 61±11 years) were included in this study and underwent whole-body FDG-PET/MRI following whole-body FDG-PET/CT. As part of the whole-body FDG-PET/MRI protocol, an EPI-sequence with 5 b-values (0, 100, 500, 1000 and 2000s/mm2) was acquired for DWI of the thorax during free-breathing. Volume of interest (VOI) measurements were performed to determine the maximum and mean standardized uptake value (SUVmax; SUVmean). A region of interest (ROI) was manually drawn around the tumor on b=0 images and then transferred to the corresponding parameter maps to assess ADCmono, Dapp and Kapp. To assess the goodness of the mathematical fit R2 was calculated for monoexponential and non-Gaussian analysis. Spearman's correlation coefficients were calculated to compare SUV values and diffusion coefficients. A Student's t-test was performed to compare the monoexponential and non-Gaussian diffusion fitting (R2). ResultsT staging was equal between FDG-PET/CT and FDG-PET/MRI in 12 of 15 patients. For NSCLC, mean ADCmono was 2.11±1.24×10−3mm2/s, Dapp was 2.46±1.29×10−3mm2/s and mean Kapp was 0.70±0.21. The non-Gaussian diffusion analysis (R2=0.98) provided a significantly better mathematical fitting to the DWI signal decay than the monoexponetial analysis (R2=0.96) (p<0.001). SUVmax and SUVmean of NSCLC was 13.5±7.6 and 7.9±4.3 for FDG-PET/MRI. ADCmono as well as Dapp exhibited a significant inverse correlation with the SUVmax (ADCmono: R=−0.67; p<0.01; Dapp: R=−0.69; p<0.01) as well as with SUVmean assessed by FDG-PET/MRI (ADCmono: R=−0.66; p<0.01; Dapp: R=−0.69; p<0.01). Furthermore, Kapp exhibited a significant correlation with SUVmax (R=0.72; p<0.05) and SUVmean as assessed by FDG-PET/MRI (R=0.71; p<0.005). ConclusionSimultaneous PET and non-Gaussian diffusion acquisitions are feasible. Non-Gaussian diffusion parameters show a good correlation with SUV and might provide additional information beyond monoexponential ADC, especially as non-Gaussian diffusion exhibits better mathematical fitting to the decay of the diffusion signal than monoexponential DWI.