Program Information
MR-Based Motion Correction in Coronary 18F-Fluoride PET Imaging Using Simultaneous PET-MR
Y Petibon1*, M Q Wilks1 , C Ma1 , C Huang2 , M T Lu3 , M K Zanni4 , S K Grinspoon4 , J Ouyang1 , G El Fakhri1 , (1) Gordon Center for Medical Imaging, Massachusetts General Hospital, Boston, MA, (2) Research Radiology and Psychiatry, Stony Brook University, Stony Brook, NY, (3) Cardiac - MR PET CT Program, Massachusetts General Hospital, Boston, MA, (4) Program in Nutritional Metabolism, Massachusetts General Hospital, Boston, MA
Presentations
TU-D-601-7 (Tuesday, August 1, 2017) 11:00 AM - 12:15 PM Room: 601
Purpose: To quantify the impact of MR-based motion correction on coronary ¹⁸F-fluoride PET imaging using simultaneous PET-MR, leveraging data from a previously published study (Zanni JID 2017).
Methods: Three HIV-infected subjects with known coronary artery disease underwent simultaneous PET-MR (Siemens mMR) for 8min, ~2hrs after ¹⁸F-fluoride injection (~150MBq). First, a 4-class attenuation map was acquired during a breath-hold using the standard MRAC sequence. List-mode PET acquisition started simultaneously with MRAC. Afterwards, thoracic motion was tracked using a navigated multi-slice spoiled gradient recalled echo sequence with a golden-angle radial trajectory (TE/TR=1.45/69.7ms, flip angle=30°, FOV=300x300x180mm³, voxel size=2.3x2.3x8mm³, 20 coronal slices). The MR navigator and recorded ECG signal were used to bin every MR radial readout and PET event into 6 cardiac and 5 respiratory phases. MR images in each cardiorespiratory phase were reconstructed from undersampled k-space data using compressed-sensing. Each MR volume was registered to a reference phase (end-exhalation/end-diastole) to calculate 4D (3D+time) motion vector fields. PET events were then reconstructed using (i) no motion correction and (ii) motion correction (MoCo). MoCo was achieved by incorporating the motion fields inside the system matrix of a modified OSEM PET reconstruction algorithm together with motion-dependent attenuation coefficients.
Results: MoCo alleviated motion-induced emission/attenuation artifacts at the lung and heart/liver interfaces in the reconstructed 18F-fluoride PET images. In total, four foci of increased coronary 18F-fluoride uptake were identified. Three of the foci co-localized with spotty calcification on antecedent coronary CT angiography. MoCo increased coronary 18F-fluoride target-to-background ratio by 21.1-46.6% (mean:30.1%) comparing with no motion correction.
Conclusion: This is the first demonstration of MR-based cardiac and respiratory motion correction in coronary ¹⁸F-fluoride PET using simultaneous PET-MR. Our results suggest that the developed methodology improves the image quality of coronary ¹⁸F-fluoride PET and has the potential to facilitate PET detection and characterization of coronary atherosclerotic plaques.
Contact Email: