Introduction

Cerebrospinal fluid (CSF) flow is known to be complex in both space and time. Currently available techniques include phase contrast method1 and time-SLIP2. The phase contrast method utilizes phase shift of the moving spins, and can be used to visualize pixelwise average flow velocity. This may become significant limitation when measuring CSF flow in small space, like the aqueduct, where large velocity difference exists even within a single voxel. The time-SLIP method, on the other hand, can visualize the net movement of water molecules between sampling periods, which is not averaged over the pixel. However, based on spin-labeling technique, the observation time is essentially limited by T1 of the tissue.Here we present a novel technique to continuously visualize slow flow, like that of CSF. It is based on SSFP sequence, and positively utilizes its dark band artifacts. When a linear gradient is added to increase field inhomogeneity along one direction, paralell dark bands appear. When the spins move, these dark bands follow the motion of the spins, and it is possible to visualize the flow or motion clearly. The response to the flow was investigated through numerical simulation. Also, the method was tested on healthy volunteers.

Methods

The sequence is shown in fig 1. It is based on balanced SSFP, with gradient pulses added to generate linear field gradients. Although it is possible to introduce the linear gradient by use of the shimming system, it is easier to control the position and direction of the dark bands when the functionarity is implemented through gradient pulses. The sequence was implemented on a 3T clinical scanner (Vantage Galan 3T, Canon Medical Systems). Numerical simulation was carried out to verify the relation between the flow and the motion of the dark bands. One-D array of voxels were considered, and SSFP response was created by solving the Bloch equation for each voxel. When the steady state was reached, the spins were moved in one direction at a constant speed. The resulting transverse magnetizaion was displayed as a 2D image, with horizontal direction as space, and the vertical direction as time. Healthy volunteers were imaged using the sequence. Images were taken continuously for about 30 seconds, at 9 frames per second. The matrix size was 96 × 192, and the FOV was 22 × 28.4cm, resulting in 2.3 × 1.5 mm in-plane resolution. Paralell imaging was used to achieve the 9 frames per second imaging rate.

Results and discussion

Results of numerical simulation is shown on fig 2. It reproduces closely what is seen in the real sequence. When the flow is slow and constant, a new steady state is reached and the position of the dark band is shifted with flow. When the flow is faster than certain limit, the dark band will follow the moving spins, just like spin labeling. Also, for slow and oscillating flow, it behaves like spin labeling again. Its behavior is a mixture of spin labeling type and phase contrast type. Fig 3 shows the results of volunteer scans. Although the motion cannot be shown on the static figure, the flow can be seen as shifts of dark bands. These bands moves following the motion of the CSF. Both slow flow and fast flow were visualized over the entire slice. Jet flow at aqueduct is also visualized, even though the intra-pixel flow velocity differences. Since the response to the flow is somewhat complex, it is difficult to measure the flow quantitatively from the image. However, it is suitable to grasp the flow pattern over a large area. The sequence is simple with just a small modification to the SSFP sequence, and requires no special resonstruction process. Also, imaging process is real time, and there is no theoretical limit to the maximum extent of imaging duration.

Conclusion

A novel technique that can visualize relatively sflow flow continously. It can be a powerful tool to investigate CSF flow.

Acknowledgements

No acknowledgement found.