Dyke award. Harmonic modulation of proton MR precessional phase by pulsatile motion: origin of spinal CSF flow phenomena.

The effects of pulsatile motion on MR imaging of spinal CSF were quantitatively evaluated with a spine phantom that simulated spinal CSF pulsation. Two fundamental interdependent pulsation flow phenomena were observed: variable reductions in signal intensity of pulsatile CSF (signal loss) and spatial mismapping of this signal beyond the confines of the subarachnoid space (phase-shift images). Phase-shift images were observed as multiple regions of signal intensity conforming morphologically to the subarachnoid space but displaced symmetrically from it along the phase-encoding axis, either added to or subtracted from stationary signal intensity. Both CSF pulsation flow phenomena occurred secondary to harmonic modulation of proton precessional phase (temporal phase shift) by the unique pulsatile motion of spinal CSF when the repetition time was not an integral multiple of the pulsation period. Each flow phenomenon was analyzed with the spine phantom independently to control individual imaging and physiologic parameters including imaging plane, repetition time, echo time, slice thickness, number of echoes, number of excitations, CSF pulsation amplitude, and CSF pulsation period. In the axial plane, signal loss was present on both first- and second-echo images and was more pronounced with larger pulsation amplitudes and smaller slice thicknesses. A quantitative relationship between these two parameters allowed the prediction of CSF pulsation amplitude when the slice thickness was known and the CSF signal intensity was measured. In the sagittal plane, signal loss was present on first-echo images, was more pronounced with larger pulsation amplitudes, and underwent incomplete even-echo rephasing on second-echo images. Phase-shift images were influenced by the relationship between repetition time and CSF pulsation period. They were partly eliminated on sagittal but not on axial second-echo images because of incomplete even-echo rephasing. Both signal loss and phase-shift images were completely eliminated with CSF gating or pseudogating, indicating the rationale for gating during clinical spinal MR. The clinical significance of these findings is that awareness of the existence of spinal CSF pulsation flow phenomena avoids diagnostic confusion, whereas understanding their etiology provides a rational approach, such as CSF gating, to eliminate them.