Child's Nervous System

, Volume 34, Issue 9, pp 1677–1682 | Cite as

A new MRI tag-based method to non-invasively visualize cerebrospinal fluid flow

  • Matthew Borzage
  • Skorn Ponrartana
  • Benita Tamrazi
  • Wende Gibbs
  • Marvin D. Nelson
  • J. Gordon McComb
  • Stefan Blüml
Multimedia Article



Abnormal cerebrospinal fluid (CSF) dynamics can produce a number of significant clinical problems to include hydrocephalus, loculated areas within the ventricles or subarachnoid spaces as well as impairment of normal CSF movement between the cranial and spinal compartments that can result in a cerebellar ectopia and hydrosyringomyelia. Thus, assessing the patency of fluid flow between adjacent CSF compartments non-invasively by magnetic resonance imaging (MRI) has definite clinical value. Our objective was to demonstrate that a novel tag-based CSF imaging methodology offers improved contrast when compared with a commercially available application.


In a prospective study, ten normal healthy adult subjects were examined on 3T magnets with time-spatial labeling inversion pulse (Time-SLIP) and a new tag-based flow technique—time static tagging and mono-contrast preservation (Time-STAMP). The image contrast was calculated for dark-untagged CSF and bright-flowing CSF. We tested the results with the D’Agostino and Pearson normality test and Friedman’s test with Dunn’s multiple comparison correction for significance. Separately 96 pediatric patients were evaluated using the Time-STAMP method.


In healthy adults, contrasts were consistently higher with Time-STAMP than Time-SLIP (p < 0.0001, in all ROI comparisons). The contrast between untagged CSF and flowing tagged CSF improved by 15 to 34%. In both healthy adults and pediatric patients, CSF flow between adjacent fluid compartments was demonstrated.


Time-STAMP provided images with higher contrast than Time-SLIP, without diminishing the ability to visualize qualitative CSF movement and between adjacent fluid compartments.


Cerebrospinal fluid (CSF) CSF flow CSF dynamics MRI Hydrocephalus ASL 


Abnormal cerebrospinal fluid (CSF) dynamics produce a number of significant clinical problems. These include hydrocephalus, loculated areas within the ventricles or subarachnoid spaces, as well as impairment of normal CSF movement between the cranial and spinal compartments that can result in cerebellar tonsillar ectopia and hydrosyringomyelia.

There are currently three MR imaging strategies for non-invasive assessment of CSF flow: (i) T2-weighted spin echo imaging, where flow is visualized as signal loss (flow-void) due to incoherent signal averaging of moving spins [1, 2]. (ii) Phase-contrast MRI that uses gradient pairs to cause quantitative changes in the phase accrual of moving spins [3]. Post-processing allows measurement of velocities and flow in regions of interest. (iii) Tag-based MRI uses regional inversion pulses, similar to arterial spin labeling, to create bright (tagged) regions of CSF. A delay between tagging and image readout allows the flow visualization of bright CSF.

Time-spatial labeling inversion pulse (Time-SLIP) is a commercially available tag-based method that has been demonstrated to work well for laminar and turbulent CSF flow at a wide range of velocities [4, 5]. However, with Time-SLIP, there are changes in the background signal due to the T1 recovery that can result in ambiguous interpretations in areas where CSF border tissue especially on images acquired with long delay times. In these cases, areas of untagged CSF may have signal intensities similar to the tagged and slow-flowing CSF.

In this study, we compared Time-SLIP with a novel tag-based methodology designed to create constant contrast between tagged and non-tagged CSF.

Materials and methods


This study included both healthy adult subjects and pediatric patients that were examined at different sites. The study of the adult subjects (five male and five female, 36 ± 18 years old) was approved by E&I Review Services (#11214), whereas the study of the pediatric patients (from neonates to teenagers) was approved by the Children’s Hospital Los Angeles Institutional Review Board (#11-00357). Informed consent was obtained for all subjects prior to studies.

MRI sequences

Studies were performed on healthy adults using a clinical 3T MR scanner (Titan 3T, Canon Medical Systems USA, Irvine, CA). A standard electrocardiography monitor with four chest electrodes and a respiratory bellows was used to monitor cardiac and respiration rates for gated sequences. We performed all pediatric patient studies on a clinical 3T MR scanner (Achieva Philips, Best, the Netherlands) with standard patient physiological monitoring.

For healthy adult subjects, both tag-based techniques used a single-shot half-Fourier fast spin echo sequence repetition time = 13,450 ms, echo time 80 ms, flip angle = 90°, acquisition matrix 224 × 224, and voxel resolution = 1.0 × 1.0 × 5.0 mm3. For Time-SLIP [5], we acquired 20 cardiac-gated images at delay times of 1700, 1800 … 3600 ms. The modified version of Time-SLIP, time static tagging and mono-contrast preservation (Time-STAMP), is identical to Time-SLIP imaging except for two modifications: (i) physiological gating was omitted and (ii) quasi-constant delay time was used (see discussion for details). For pediatric patients, Time-STAMP sequences were implemented with repetition time = 8000 ms, echo time 7 ms, flip angle = 70°, acquisition matrix 256 × 161, voxel resolution = 0.9 × 1.4 × 4.0 mm3, and typically nine ungated images with a delay time of 2700 ms. Time-STAMP sequence differences and Time-SLIP omission in these patients were due to scanner vendor issues.

Image processing

All images acquired from adults were analyzed with MATLAB. For contrast assessments in these subjects, ROIs of dark-untagged CSF and bright-flowing CSF were selected and the mean and standard deviation were calculated and voxels that exceeded two standard deviations of the mean were removed.


We assessed differences in contrast in the adult subjects first via D’Agostino and Pearson normality test. We determined a non-parametric test was appropriate so we used Friedman’s test with Dunn’s correction for multiple comparisons. Mean values and standard deviations are reported.


The image contrasts were consistently superior on Time-STAMP images when compared with Time-SLIP images, as determined by both ROI analysis (Fig. 1) and by inspection (Fig. 2; best viewed as a video, in supplementary materials online).
Fig. 1

Dynamic images for Time-SLIP (top) and Time-STAMP (bottom) acquisitions in the same subject. Time-SLIP images acquired with delays from 1700 to 3600 ms. For Time-STAMP, a constant delay time of 2500 ms was used. Both series demonstrate the same CSF movements. Tagged CSF flows out of the third and into the fourth ventricles, oscillates in the prepontine cistern, and flows into the fourth. However, the poor contrast in Time-SLIP makes it more difficult to observe the flow than in the respective Time-STAMP images. The omission of cardiac synchronization did not impair the ability to observe flow patterns in any of our subjects as qualitatively assessed by the authors

Fig. 2

Top: Healthy adult Time-SLIP intensities for regions of interest of the images shown in Fig. 1. Note the considerable variation in the Time-SLIP untagged dark CSF (blue) should be compared with those below. Below: same healthy adult, Time-STAMP intensities for regions of interest for the images shown in Fig. 2. Note the lack of variation in the untagged dark CSF (blue) that is making it more readily distinguishable from the untagged CSF (black)

For each adult subject, the contrasts were analyzed for comparison (Table 1). We found that Time-STAMP had significantly better contrast for all comparisons (p < 0.0001). In the pediatric population, Time-STAMP images were acquired for CSF spaces as clinically indicated: lateral, third, and fourth ventricles, pre-pontine and interpeduncular cistern and cisterna magna. Time-SLIP was unavailable on our clinical scanner so direct comparisons could not be conducted. However, we reviewed the Time-STAMP images with the same qualitative basis and observed the same bulk and turbulent flows as have been described with Time-SLIP [5]. Similar dynamics of CSF flow are readily apparent in the Time-STAMP images of patients (Figs. 3, 4).
Table 1

Contrast ratios of Time-STAMP and Time-SLIP intensities

Subject ID

Flowing CSF vs dark CSF



























Standard deviation


CSF cerebrospinal fluid

Fig. 3

Pediatric patient Time-STAMP intensities for regions of interest. This patient is an 18-month-old male with hydrocephalus secondary to intraventricular hemorrhage as a premature neonate. Limited MRI with Time-STAMP allows monitoring of ventricle size and function of ETV. Note that the intensities show similar patterns as Time-STAMP in a healthy adult (Fig. 2)

Fig. 4

This 16-year-old female had a very complicated course of treatment for hydrocephalus that took well over 2 years to stabilize. She initially had a ventriculoperitoneal shunt placed for progressive hydrocephalus but developed ascites. Multiple CSF aerobic, anaerobic, and fungal cultures were sterile. Next was the placement of a ventriculopleural shunt, but she developed a pleural effusion. It was evident that she had a fourth ventricular outflow obstruction. All CSF cultures continued to be sterile. The fourth ventricle was fenestrated via a posterior fossa craniotomy. It was patent for a while but then sealed again. Finally, a CSF culture was positive for histoplasmosis. She was started on antifungal medication that would more likely cure the infection if no shunt hardware was present. Time-STAMP study showed a dilated ventricular system with fourth ventricle outflow obstruction but flow through the aqueduct, basal cisterns, and foramen of Monro (a). An ETV was done and the Time-STAMP imaging demonstrated the presence of robust flow through the ETV (b). Blocked outflow from the fourth ventricle remained, but the ETV was successful in controlling the hydrocephalus (c). This patient is doing well on continued fungal treatment


In this study, we compared two tag-based CSF flow visualizations. The first—Time-SLIP—is commercially available. Time-SLIP uses cardiac or respiratory gating to trigger a radio frequency pulse to selectively “tag” a region of interest (bright). After a delay time, an image is obtained with a fast single-shot acquisition. The delay time allows the tracing of CSF as it moves from one compartment to another. With the acquisition of a series of images with different delay times, the dynamics of CSF flow can be captured and visualized. An inherent disadvantage of this method is that the contrast between tagged, bright CSF, and untagged regions (both CSF and parenchyma) is a function of the delay time. Particularly at long delay times, T1-recovery of untagged regions results in signal intensities that are comparable with those of the tagged CSF, leading to ambiguous interpretations. To address this ambiguity, we tested a modified approach (Time-STAMP) where the contrast is constant.

To objectively compare the performance of the two tagged-based methods, we evaluated contrasts between tagged CSF, untagged CSF, and parenchyma in ten adult volunteers with Time-SLIP and Time-STAMP. All acquisition parameters were identical except varying delay times for Time-SLIP versus constant delay times for Time-STAMP and gating disabled for Time-STAMP. We used two methods to measure the contrast: comparing individual image contrast and summing contrast over a series of images. Both approaches consistently demonstrated improved contrast with Time-STAMP.

To assess the feasibility of Time-STAMP for CSF imaging in patients, we evaluated 96 pediatric patients with a variety of etiologies, and used Time-STAMP to assess flow in a variety of CSF spaces and endoscopic third ventriculostomy (ETV) patency. Two case studies (Fig. 3 and Fig. 4) follow.

The assessment of CSF flow is clinically important to verify flow through the aqueduct of Sylvius, foramen of Monro, stoma of endoscopic third ventriculostomy, fenestration of arachnoid cysts and loculated CSF spaces, craniocervical junction, and CSF collections anywhere in the central nervous system where the isolation of a CSF compartment or success of a fenestration is unknown.

For the majority of these clinical conditions, the quantitation of CSF (i.e., measuring velocities and flow rates), while of interest, is clinically less relevant than verifying qualitative flow between two compartments that is present or has been restored after an intervention. Clinically, it is also important that tag-based methods have acquisition times of 3–5 min, and can label CSF at any orientation and at any place in the central nervous system. Finally, when scanning patients, particularly unsedated pediatric patients, head movements can cause problems with flow-void MRI and phase-contrast MRI, both of which require signal sampling over several minutes and are thus prone to gross motion artifacts. This is not the case with tag-based methods with fast single-shot image readouts.

This non-invasive MRI method uses CSF itself as the tracer. These studies are readily repeatable for continuing evaluation of interventions be it under normal physiological conditions or in altered states.


Time-STAMP provides superior contrast compared to Time-SLIP, omits the requirement for cardiac or respiratory gating, and adds but little to scan time (2–3 min). Also, Time-STAMP produces diagnostic images despite patient movement, thus eliminating the need for sedation.



This manuscript is published in memoriam of William G. Bradley: a distinguished clinician and researcher. We would also like to thank JF and DB at Canon Medical Systems USA for their help recruiting volunteers and providing imaging time.

Funding information

This work was made possible by the financial support of the Rudi Schulte Research Institute.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Supplementary material

Video 1

Left: dynamic images for Time-SLIP in a healthy adult subject. The poor contrast in the third ventricle makes it more difficult to observe the flow that may be present. Right: dynamic images for Time-STAMP in a healthy adult subject. The omission of cardiac synchronization did not impair the ability to observe flow patterns, but the use of a static delay time improved the contrast. Improvements to contrast make it easier to observe flow (MP4 3429 kb)

Video 2

Dynamic images for Time-STAMP in the same 16-year-old shown in Fig. 4. Flow from the fourth ventricle through the patent aqueduct of Sylvius to the 3rd ventricle, as well as flow at the craniocervical junction and prepontine cisterns indicate ETV is likely to be successful (A). Post ETV shows good communication of the fourth ventricle, aqueduct of Sylvius, and third ventricle within the prepontine cisterns (B). Post ETV shows no communication of the outlet of the fourth ventricle, and the presence of turbulent flow within the fourth ventricle (C) (M4V 147 kb)


  1. 1.
    Bradley WG, Kortman KE, Burgoyne B (1986) Flowing cerebrospinal fluid in normal and hydrocephalic states: appearance on MR images. Radiology 159:611–616CrossRefPubMedGoogle Scholar
  2. 2.
    Bradley WG, Whittemore AR, Kortman KE, Watanabe AS, Homyak M, Teresi LM et al (1991) Marked cerebrospinal fluid void: indicator of successful shunt in patients with suspected normal-pressure hydrocephalus. Radiology 178:459–466CrossRefPubMedGoogle Scholar
  3. 3.
    Sharma AK, Gaikwad S, Gupta V, Garg A, Mishra NK (2008) Measurement of peak CSF flow velocity at cerebral aqueduct, before and after lumbar CSF drainage, by use of phase-contrast MRI: utility in the management of idiopathic normal pressure hydrocephalus. Clin Neurol Neurosurg 110:363–368CrossRefPubMedGoogle Scholar
  4. 4.
    Yamada S, Tsuchiya K, Bradley WG, Law M, Winkler ML, Borzage MT et al (2014) Current and emerging MR imaging techniques for the diagnosis and management of CSF flow disorders: a review of phase-contrast and time-spatial labeling inversion pulse. Am J Neuroradiol 36(4):623–30.
  5. 5.
    Yamada S, Miyazaki M, Kanazawa H, Higashi M, Morohoshi Y, Bluml S, McComb JG (2008) Visualization of cerebrospinal fluid movement with spin labeling at MR imaging: preliminary results in normal and pathophysiologic conditions. Radiology 249:644–652CrossRefPubMedGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  • Matthew Borzage
    • 1
    • 2
    • 3
  • Skorn Ponrartana
    • 2
    • 3
    • 4
  • Benita Tamrazi
    • 2
    • 4
  • Wende Gibbs
    • 4
  • Marvin D. Nelson
    • 2
  • J. Gordon McComb
    • 5
    • 6
  • Stefan Blüml
    • 2
    • 3
  1. 1.Fetal and Neonatal Institute, Division of NeonatologyChildren’s Hospital Los AngelesLos AngelesUSA
  2. 2.Department of RadiologyChildren’s Hospital Los AngelesLos AngelesUSA
  3. 3.Rudi Schulte Research InstituteSanta BarbaraUSA
  4. 4.Department of Radiology, Keck School of MedicineUniversity of Southern CaliforniaLos AngelesUSA
  5. 5.Division of NeurosurgeryChildren’s Hospital Los AngelesLos AngelesUSA
  6. 6.Department of Neurological Surgery, Keck School of MedicineUniversity of Southern CaliforniaLos AngelesUSA

Personalised recommendations