Christopher M. Rank MSc, Presenter: Nothing to Disclose

Thorsten Heusser DIPLPHYS, Abstract Co-Author: Nothing to Disclose

Marcus Brehm, Abstract Co-Author: Nothing to Disclose

Thomas Beyer PhD, Abstract Co-Author: Nothing to Disclose

Marc Kachelriess PhD, Abstract Co-Author: Nothing to Disclose

To compensate for respiratory patient motion in PET/MR using information from a strongly undersampled radial MR sequence that a) runs in parallel with the PET acquisition, that b) can be interlaced with other MR sequences, and that c) requires less than one minute of the total MR acquisition time per bed position.

We use a 2D radial stack-of-stars sampling scheme to continuously monitor patient motion during PET/MR acquisitions. Each slice is sampled with a rate of 160 radial spokes/min to achieve 16 radial spokes spread across a respiratory phase. Assuming 100 slices and a repetition time of 3.6 ms, our sampling scheme takes less than one minute of the MR scan time per bed position. Based on the gated but undersampling artifact-contaminated 4D MR images, motion vector fields (MVFs) were estimated using our newly-developed artifact model-based deformable registration [Med Phys 39:7603, Med Phys 40:101913] to obtain high fidelity 4D MR images and MVFs. Gated or list mode PET images were reconstructed using a MoCo 3D OSEM algorithm based on these MVFs. We simulated a 4D PET/MR image volume of the breathing thorax (6.0 kBq/mL soft tissue activity) based on a static MR volume of a patient with two hot lung lesions (8 and 16 mm spheres, 25 kBq/mL activity). The simulation corresponds to the Biograph mMR system. For quantification, SUVmean values of both artificial lesions were calculated and compared to a reference gated 4D PET reconstruction with ten-fold measurement time.  

Visual inspection of the PET images showed that the small lesion was well detected on the MoCo 4D images but detectability was diminished on the 3D and gated 4D reconstructions. SUVmean of the 8 mm and 16 mm lesion was 0.80 and 1.92 (3D), 1.58 and 2.96 (gated 4D), 1.23 and 3.50 (MoCo 4D) and 1.45 and 3.68 (reference gated 4D), respectively.

In this simulation study, MR-based MoCo of PET images acquired during normal respiration yielded a significant visual and quantitative improvement compared to 3D and gated 4D reconstructions for the same acquisition time.

The proposed MoCo method could be potentially integrated into clinical PET/MR to improve PET quantification and image quality, thus, increasing the diagnostic value of PET/MR.

Rank, C, Heusser, T, Brehm, M, Beyer, T, Kachelriess, M, Motion Compensation (MoCo) for Simultaneous PET/MR Based on Strongly Undersampled Radial MR Data – A Simulation Study.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14004132.html