PET motion correction using simultaneous PET-MR acquisition and MR-derived motion fields

Charalampos Tsoumpas, Jane E. Mackewn, Andrew King, Christian Buerger, John Totman, Tobias Schaeffter, Paul Marsden

Research output: Contribution to journalMeeting abstract

Abstract

Objectives: Subject motion is one of the main factors limiting PET performance. In this study we demonstrate the use of simultaneous PET-MR to motion-correct PET images using motion fields estimated from simultaneously acquired MR data. Methods: PET and MR data were acquired with a small MR-compatible PET system (Mackewn et al. 2008) and a 3T clinical MR (Achieva, Philips Healthcare) with a standard receive-only head-coil. Phantom experiments were performed to investigate corrections for rigid and non-rigid motion. A hot-rod resolution phantom (OD:3cm) was filled uniformly with water and 18F, and rotated in discrete steps. PET-MR data were reconstructed at each position. To simulate non-rigid motion three stationary line sources were added adjacent to the rotating phantom. Rigid and non-rigid MR deformation fields were obtained by a normalised mutual information intensity-based registration (Rueckert et al. 1999) of each position with the reference. The sequential PET images were i) added to create an uncorrected image; ii) transformed and added to create a motion-corrected image. A stationary PET image with similar total counts was also acquired. Results: The uncorrected PET image showed clear artefacts and loss of resolution, whilst individual stationary frames were less blurred but had poor SNR. Both rigid and non-rigid corrected image show significant reduction in artefacts and increased SNR, and are comparable with the images acquired with no motion. Conclusions: We used simultaneously acquired PET-MR data to demonstrate that MR-derived motion fields can be used to motion-correct dynamic PET images. Research Support: The research leading to this project has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under the grant agreement No: 201651.
Original languageEnglish
Pages (from-to)532
Number of pages1
JournalJournal of Nuclear Medicine
Volume50
Publication statusPublished - 2009
Event2009 SNM Annual Meeting - Toronto, Canada
Duration: 1 Jan 2009 → …

Fingerprint

Dive into the research topics of 'PET motion correction using simultaneous PET-MR acquisition and MR-derived motion fields'. Together they form a unique fingerprint.

Cite this