RT Journal Article SR Electronic T1 Data-driven motion compensation in PET using a MRI VIBE 3D GRE sequences JF Journal of Nuclear Medicine JO J Nucl Med FD Society of Nuclear Medicine SP 1971 OP 1971 VO 57 IS supplement 2 A1 Martin Lyngby Lassen A1 Lucian Beer A1 Ivo Rausch A1 Helmut Prosch A1 Inki Hong A1 Michael Casey A1 Matthias Fenchel A1 Thomas Beyer YR 2016 UL http://jnm.snmjournals.org/content/57/supplement_2/1971.abstract AB 1971Objectives To investigate the effect of respiratory motion compensation based on a prototype 3D radial VIBE GRE MRI sequence in an integrated whole-body PET/MRI system (Biograph mMR, Siemens).Methods Six patients with lesions in the lungs (1/6) and mediastinum/lung wall (5/6) were analyzed using 18F-FDG-PET (163±10 MBq, 40±2min p.i.). In all patients MRI and PET data were acquired simultaneously for 10-min using a free-breathing protocol. Data-driven respiratory motion compensation was based on respiratory signals extracted from a 3D VIBE GRE sequence. The emission data was binned into 5 uniform sinograms encompassing data from the end-exhale to the end-inhale position of each respiratory cycle. All sinograms were co-registered to the first sinogram bin, containing data of the most frequent respiratory amplitude. The effect of motion compensation was assessed through three reconstructions: a motion-compensated image (MoCo), a static reconstruction (Stat) and a single-phase reconstruction (SBin) using a prototype reconstruction software. We report changes in the SUVpeak, SUVmean and lesion volume (based on the threshold activity of the 42% iso-contour of Stat). Lesion-to-liver ratios (LLR: 42% SUVmax/mean SUV for a 3cm ø-sphere in the liver) were used to assess changes in the lesion-to-background of the reconstructed PET images.Results LLR for the patients was 5.7±3.4 (stat), 6.1±4.3 (MoCo) and 6.9±4.6 (SBin), with statistically difference between (Stat:SBin, p=0.03). No statistically difference were found between (MoCo:SBin) and (Stat:MoCo), p>0.05. Average SUVpeak of the lung lesions was 6.6±3.1 (Stat), 6.6±2.9 (MoCo) and 6.4±3.1 (Sbin), with p>0.05. Minor increases in the corresponding SUVmean were seen for MoCo and SBin compared to Stat: 5.0±2.1 (Stat), 5.2±2.5 (MoCo) and 6.0±3.6 (Sbin) with p>0.05. Lesion volumes (42% iso-contour threshold of Stat) was constant for all reconstructions; 31.2±42.2mL (Stat), 31.3±42.7mL (MoCo) and 28.4±36.9mL (SBin) with p>0.05.Conclusions A minor increase (p>0.05) in the SUVmean was observed for lung lesions when compensating for the respiratory motion (MoCo and SBin) in fully-integrated PET/MR. LLR varied with the type of reconstruction with LLR increasing for motion-compensated images. Further improvements of the motion compensation are expected when applying the MoCo on patients with lesions more vulnerable to respiratory motion.