The cerebrospinal fluid (CSF) flow in the cervical spine is expected to be reduced in persons with spinal cord injury (SCI) as a result of an increased resistance to the flow from the stenosis in the spinal subarachnoid space (SSAS) and a reduced stroke volume of the heart.¹ Therefore, the CSF flow in the SSAS at the cervical spine was compared between healthy controls and individuals with SCI using phase contrast MRI.The flow velocity in the cranial-caudal direction was measured at the C4 spine level of 10 healthy control and 10 SCI participants (4 AIS A, 4 AIS B, 2 AIS C) using a retrospectively gated phase-contrast cine sequence with a peripheral pulse obtained from a finger and a velocity encoding range of 10 cm/s at 3 T MRI with a head-and-neck RF coil. The region of detectable flow in SSAS was segmented semi automatically using a spectral decomposition of the cine images of a complex data format (Fig. 1).² The spinal vessels adjacent to the SSAS were removed using the expected property that the volume flow rate averaged over the cardiac cycle in the spinal vessels was greater than that of CSF in the SSAS. The baseline velocity offset in the flow region was corrected automatically using a property of smooth spatial variation of the phase offset in the image plane.³ The flow in the segmented SSAS was analyzed for flow parameters such as SSAS cross-section area, cyclic flow rate (summation of flow rate during one cardiac cycle), and velocities. The extracted flow parameters were statistically compared between the two populations using a Mann-Whitney U-test. Additionally, the flow parameters of all participants were analyzed for correlation between the SSAS cross-section area, cardiac cycle length, and velocities using a linear regression analysis.The velocity offset and systolic/diastolic velocity maps of a healthy volunteer are shown in Fig. 2. The velocity offset was spatially inhomogeneous and the velocities were not homogeneous within SSAS either. The cardiac cycle length was longer in SCI participants than in control subjects. Accordingly, the time to the systolic velocity was delayed in SCI participants (Table 1). It is notable that the SSAS cross-section area was smaller, and both systolic and diastolic velocities were faster in SCI participants, and that the cyclic flow rate was significantly smaller in SCI participants than in control subjects. With the exception of the cardiac cycle length, all the above mentioned differences were statistically significant. In linear regression analyses of flow parameters of all subjects in the two populations, the systolic time delay and systolic velocity were proportional to the cardiac cycle length with a statistical significance (Fig. 3). On the other hand, the systolic time delay, systolic velocity and diastolic velocities were inversely proportional to the SSAS cross-section area with a statistical significance (Fig. 4).The flow velocity of SCI participants was faster than that of healthy volunteers in both systolic and diastolic cardiac phases. This finding is contrary to the expected flow velocities given that a slower heart rate and a reduced ejection fraction and stroke volume of the heart are often seen in SCI participants. Although the decreased heart rate did not correlate with CSF flow velocity, the decreased SSAS cross-section area did correlate with increased CSF flow. Considering fluid dynamics, decreasing the area of flow should increase the velocity of fluid flow. Therefore, the reduced SSAS cross-section area may be a contributing factor to the increased velocity of CSF flow within SSAS in SCI participants.