An improved MR sequence for attenuation correction in PET/MR hybrid imaging.

The aim of this study was to investigate the effects of MR parameters on tissue segmentation and determine the optimal MR sequence for attenuation correction in PET/MR hybrid imaging. Eight healthy volunteers were examined using a PET/MR hybrid scanner with six three-dimensional turbo-field-echo sequences for attenuation correction by modifying the echo time, k-space trajectory in the phase-encoding direction, and image contrast. MR images for attenuation correction were obtained from six MR sequences in each session; each volunteer underwent four sessions. Two radiologists assessed the attenuation correction maps generated from the MR images with respect to segmentation errors and ghost artifacts on a five-point scale, and the scores were decided by consensus. Segmentation accuracy and reproducibility were compared. Multiple regression analysis was performed to determine the effects of each MR parameter. The two three-dimensional turbo-field-echo sequences with an in-phase echo time and radial k-space sampling showed the highest total scores for segmentation accuracy, with a high reproducibility. In multiple regression analysis, the score with the shortest echo time (-3.44, P<0.0001) and Cartesian sampling in the anterior/posterior phase-encoding direction (-2.72, P=0.002) was significantly lower than that with in-phase echo time and Cartesian sampling in the right/left phase-encoding direction. Radial k-space sampling provided a significantly higher score (+5.08, P<0.0001) compared with Cartesian sampling. Furthermore, radial sampling improved intrasubject variations in the segmentation score (-8.28%, P=0.002). Image contrast had no significant effect on the total score or reproducibility. These results suggest that three-dimensional turbo-field-echo MR sequences with an in-phase echo time and radial k-space sampling provide improved MR-based attenuation correction maps.