1357523 - Whole-heart non-rigid motion-compensated 3D T2 mapping and 18FDG PET in a PET/MR system

Combining several investigations, including LGE CMR and 18F-FDG PET in addition to histology¹ has been suggested for cardiac sarcoidosis (CS) diagnosis. Simultaneous 18F-FDG PET and quantitative 2D T₂-mapping has shown promise for improved diagnosis, however misalignments remain a challenge. A free-breathing motion-corrected 3D whole-heart T₂-mapping sequence acquired simultaneously with 18F-FDG PET at a 3T PET-MR system was proposed here. This approach enables the non-rigid motion-correction for both the 3D T₂-mapping and the PET data to the same respiratory position, resulting in aligned volumes for improved clinical interpretation. In this study, we tested this approach in two patients with suspected cardiac sarcoidosis. 

Methods:

3D T₂-mapping sequence (T₂-prep=0, 28, 55ms) was acquired simultaneously with list-mode PET data on a 3T PET-MR system (Biograph mMR, Siemens) (Fig.1). A 3-fold undersampled variable-density Cartesian trajectory² is used. The sequence includes a saturation pulse, a fat saturation pulse and image navigators (iNAVs)³. Translational (TR) respiratory motion is estimated from 2D iNAVs and virtual 3D (v3D) iNAV based on autofocus^4,5. TR motion is used to correct and bin the MR data and produce 3D images. 3D non-rigid motion is then estimated and incorporated into a multi-contrast motion-corrected (MC) MR reconstruction^6,7. T₂-maps are computed using dictionary-matching⁸. TR motion from MR data is used to also bin the PET data. The μ-map is registered to the third 3D T₂-map contrast to improve alignment between the μ-map and PET image position. The third 3D T₂-map contrast is used together with the estimated non-rigid motion fields to perform MR-guided MC PET reconstruction⁹. 3D MR and PET images are corrected to the same respiratory position, enabling direct fusion of both datasets. We tested this approach in two patients with suspected      cardiac sarcoidosis. MR imaging parameters: coronal orientation, FOV=336x336x156-160mm³, 1.5 mm³ isotropic resolution, TR/TE=3.45/1.57ms, FA=15°, scan time~12 min. Proposed technique was compared against conventional 2D T₂ mapping (Fig.2).

Results:

T₂ values of the proposed 3D T₂-mapping sequence were slightly underestimated compared with the conventional 2D T₂-mapping, due to fewer slices (3 vs ~50) analysed with the conventional 2D T₂ map compared to the 3D T₂ map (Fig.1). Image quality in the PET data visually improved due to motion correction as well as it improved correspondence to the 3D MR data compared to conventional uncorrected PET reconstructions (Fig.3). No T₂ elevation or 18F-FDG uptake was observed in these subjects.

Conclusion:

A novel approach for simultaneously acquired 18F-FDG PET and 3D T₂-mapping data was successfully evaluated in two patients with suspected cardiac sarcoidosis. 3D T₂ values were comparable with the conventional 2D T₂ values. Improved PET images due to motion correction and MR guidance call for future studies in more patients with cardiac sarcoidosis undergoing simultaneous PET-MR.