Respiratory synchronized MR image generation using 4D PET acquisitions

International audienceObjectives: Respiratory motion correction methodologies in combined PET/MR are almost exclusively based on the use of respiratory synchronized MR acquisitions to derive deformable transformations for motion correction in PET. On the other hand, the acquisition of 4D MR datasets is not currently common clinical practice. The objective of this work is therefore to generate 4D MR images from a single static MR image and deformation matrices obtained from simultaneously acquired 4D non-attenuation corrected (NAC) PET images. This approach eliminates the need for 4D MR acquisitions, facilitating accurate MR based attenuation correction (AC) in simultaneously acquired 4D PET. Methods: Two patient datasets acquired on the SIEMENS mMR PET/MR system were used. 4D PET acquired datasets were retrospectively binned into 4 motion amplitude frames corresponding to simultaneously acquired 4D gradient echo MR images. A single MR Dixon scan at end expiration was also acquired for PET AC purposes. The reconstructed 4D NAC PET images were registered using a deformable registration algorithm to the end expiration NAC PET image. The derived deformation fields were subsequently applied to the end expiration 3D gradient echo MR volume to obtain respiratory synchronized MR images. The accuracy of the proposed method was assessed by comparing the generated to the original 4D MRI images using a global correlation coefficient and expert validation by identifying anatomical landmarks in both 4D MRI image series. Results: High correlation coefficient (0.91± 0.03) and small errors (