Tag-based CSF flow imaging methods are variations of arterial spin labeling sequences designed to work well for imaging CSF flow. These methods are characterized by their approach, wherein they tag (or label) one region of CSF, and readout the image in another region or orientation. The tag is commonly a double inversion, wherein the entire brain receives one excitation tag, and then a region of interest is selected and labeled with a second excitation tag. Thus, the untagged-CSF in the brain will undergo a T1 inversion recovery, whereas the tagged CSF is fully-recovered after the pair of tags.
One common use is to label the CSF in the fourth ventricle, but use a midline sagittal readout to detect flow through either the cerebral aqueduct of Sylvius or foramen of Magendie. The principal considerations when designing a tag-based flow acquisition sequence are the suitability of the anatomy and physiology of the flow for a tag-based imaging approach.
Tag-based flow sequences generally work best when one large volume of CSF (typically a ventricle or cyst), is separated from another large volume of CSF by a narrow aqueduct or fenestration. In the overall approach involves tagging one CSF volume and imaging the other volume. This approach will either demonstrate tagged CSF flowed (or did not flow) between volumes, thereby indicating if the structure are patent (or not). Thus, both normal ventricular anatomy and patients undergoing evaluation for hydrocephalous are highly amenable to tag-based CSF flow imaging.
Very small (i.e. single or sub-voxel) sized structures are not generally suitable for imaging with tab-based imaging methods because tagged CSF that flows into these structures will be partially-volumed along with parenchyma.
Understanding what physiology is suitable is best done by first understanding the mechanism of the MRI tag. This tag uses a double-inversion, one inversion applied to the whole brain (untagged CSF), and a second inversion applied only to the region of interest (tagged CSF). The ideal time to distinguish the tagged and untagged CSF is when the intensity of the untagged CSF signal is zero and the intensity of the tagged CSF is at its maximum.
Because the tagged CSF is inverted, then immediately inverted again, it will have its full intensity throughout the entire MRI sequence. However, the untagged CSF undergoes an inversion recovery; it starts with a non-zero (negative) intensity, and as it recovers it becomes zero, and then eventually non-zero (positive). The zero-point in the untagged CSF happens approximately 2630 ms after the excitations (assuming T1 property of CSF is approximately 3800 ms). Thus, the ideal flows are those which occur over roughly this speed.
Some biological flow phenomena are highly predictable. In these cases Tag-based CSF flow sequences can be advantage of imaging features that average together data from multiple individual acquisitions (multi-shot or signal averaging) to improve image quality. If flow is periodic, e.g. driven by heart beats or breaths then a cardiac or respiratory gated sequence would be needed. Alternatively, if flow is constant then no gating would be required.
Other biological flow phenomena are unpredictable. In these cases, Tag-based CSF flow should take advantage of features that allow entire images to be derived from single acquisitions (single-shot). These resulting images will often be lower quality (versus those with multi-shot or signal averaging).
Compared to phase contrast flow imaging, tag-based CSF flow imaging can take use relatively larger voxels (lower in-plane resolution) and thicker slices (lower thorough-plane resolution) without major problems from partial voluming. This is because the tag-based CSF flow imaging does not need to image small structures (e.g. cerebral aqueduct).