T2*-weighted imaging of the spinal cord suffers from signal dropouts that hamper blood-oxygenation-level-dependent functional magnetic resonance imaging (fMRI). They are due to field inhomogeneities caused by the different magnetic susceptibilities of the vertebrae and the intervertebral disks that vary periodically along the cord and, thus, cannot be compensated appropriately with conventional (constant) shimming. In this study, a single, slice-specific gradient pulse ("z-shim") is applied in echo-planar imaging of axial sections in order to compensate for the corresponding through-slice signal dephasing without affecting the acquisition time, i.e. the temporal resolution. Based on a reference acquisition sampling a range of compensation moments, the value yielding the maximum signal amplitude within the spinal cord is determined for each slice. Severe N/2 ghosting for larger compensation moments is avoided by applying the gradient pulse after the corresponding reference echoes. Furthermore, first-order flow compensation in the slice direction of both the slice-selection and the z-shim gradient pulse considerably reduces signal fluctuations in the cerebro-spinal fluid surrounding the spinal cord, i.e. would minimize ringing artifacts in fMRI. Phantom and in vivo experiments show the necessity to use slice-specific compensation moments in the presence of local susceptibility differences. Measurements performed in a group of 24 healthy volunteers at 3T demonstrate that this approach improves T2*-weighted imaging of axial sections of the cervical spinal cord by (i) increasing the signal intensity (overall by about 20%) and (ii) reducing signal intensity variations along the cord (by about 80%). Thus, it may help to improve the feasibility and reliability of fMRI of the spinal cord.
T2*-weighted imaging of the spinal cord suffers from signal dropouts that hamper blood-oxygenation-level-dependent functional magnetic resonance imaging (fMRI). They are due to field inhomogeneities caused by the different magnetic susceptibilities of the vertebrae and the intervertebral disks that vary periodically along the cord and, thus, cannot be compensated appropriately with conventional (constant) shimming. In this study, a single, slice-specific gradient pulse ("z-shim") is applied in echo-planar imaging of axial sections in order to compensate for the corresponding through-slice signal dephasing without affecting the acquisition time, i.e. the temporal resolution. Based on a reference acquisition sampling a range of compensation moments, the value yielding the maximum signal amplitude within the spinal cord is determined for each slice. Severe N/2 ghosting for larger compensation moments is avoided by applying the gradient pulse after the corresponding reference echoes. Furthermore, first-order flow compensation in the slice direction of both the slice-selection and the z-shim gradient pulse considerably reduces signal fluctuations in the cerebro-spinal fluid surrounding the spinal cord, i.e. would minimize ringing artifacts in fMRI. Phantom and in vivo experiments show the necessity to use slice-specific compensation moments in the presence of local susceptibility differences. Measurements performed in a group of 24 healthy volunteers at 3T demonstrate that this approach improves T2*-weighted imaging of axial sections of the cervical spinal cord by (i) increasing the signal intensity (overall by about 20%) and (ii) reducing signal intensity variations along the cord (by about 80%). Thus, it may help to improve the feasibility and reliability of fMRI of the spinal cord.