Time-Spatial Labeling Inversion Pulse (Time-SLIP) MRI for Evaluating Cerebrospinal Fluid Velocity and Visualizing Flow Dynamics in Patients With Chiari Type I Malformation.

Background and objectives: Phase-contrast MRI is unstable and is not widely implemented in the imaging of Chiari malformation type I (CM-I) because of its low signal-to-noise ratio and the need for subsequent additional averaging. Time-spatial labeling inversion pulse MRI (T-SLIP MRI) is an emerging imaging modality with a high signal-to-noise ratio. This study is the first to examine cerebrospinal fluid (CSF) dynamics on the basis of velocity exclusively in patients with CM-I using T-SLIP MRI before and after posterior fossa decompression.

Methods: Eleven patients with CM-I underwent T-SLIP MRI before and/or after posterior fossa decompression. CSF dynamics were analyzed at 5 points around the craniovertebral junction. T-SLIP measurements included (1) targeted CSF labeling; (2) manual frame-by-frame annotation of the labeled CSF wave; (3) description of CSF flow in terms of wave functions calculated using computation software; and use of this function for (4) calculation of CSF velocity (rostral and caudal peak), total distance traveled by labeled CSF, and mean CSF velocity ( $\overline{V}$ ). Differences between preoperative and postoperative peak velocity (rostral and caudal) and $\overline{V}$ were assessed using paired t-test.

Results: Rostral and caudal peaks significantly increased at 2 of the 5 points (40%), whereas $\overline{V}$ significantly increased at 4 points (80%), altogether covering all observation points with significant changes. CSF filling the syrinx through the syrinx wall from the spinal subarachnoid space and complex CSF flow at the dorsal craniovertebral junction were captured preoperatively and postoperatively, respectively.

Conclusion: T-SLIP MRI data for patients with CM-I were successfully quantified on the basis of velocity. Tailor-made optimal decompression should be pursued based on both T-SLIP data with high accuracy and bibliographical craniometric data with surgical outcomes, which can now be easily and comprehensively analyzed using machine learning.