Quantitative evaluation of normal cerebrospinal fluid flow in Sylvian aqueduct and perivascular spaces of middle cerebral artery and circle of Willis using 2D phase-contrast MRI imaging

Introduction: Recently, it was proposed that CSF flow comprises a critical part of the glymphatic system, playing a role in various brain abnormalities from Alzheimer’s disease to hydrocephalus. Thus, CSF flow measurements have been increasingly used for diagnostic and clinical monitoring purposes. However, CSF flow in the periarterial spaces of the circle of Willis and the middle cerebral artery remain unexplored.

Methods: We employed phase-contrast MRI to establish baseline parameters of CSF flow along the perivascular spaces of the circle of Willis and the middle cerebral artery and compare them with the Sylvian aqueduct. We also developed a new, semi-automated method for outlining the perivascular spaces and extracting CSF flow parameters. The 24 healthy participants were recruited to achieve an even distribution by age (mean: 40 ± 11) and gender (13 males, 11 females).

Results: For most CSF flow parameters, the circle of Willis and middle cerebral artery were similar but differed from the Sylvian aqueduct. The linear mixed models and general linear mixed models for CSF flow parameters, except for time to peak velocity, indicated strong effects of the conduits. CSF velocity was lower by 0.159 cm/s in the circle of Willis and 0.198 cm/s in the middle cerebral artery than in the Sylvian aqueduct. Overall, differences in CSF flow parameters between sex and age groups were negligible.

Discussion: Our semi-automated routine for CSF flow measurements in the Sylvian aqueduct (0.00700 mL/s) aligned with the range of literature values, 0.0049–0.0432 mL/s. In this study, we have established baseline values of CSF flow along the circle of Willis and the middle cerebral artery as well as highlighted the limited influence of sex and/or age.

1 Introduction

Cerebrospinal fluid (CSF) is a clear fluid surrounding the central nervous system and filling the cerebral ventricles and the central canal of the spinal cord. At any given time, there are 90–150 mL of CSF in the cranium and spinal cord because production and absorption occur at the same rate to maintain intracranial pressure (Sartoretti et al., 2019; Vandenbulcke et al., 2022). CSF is produced in the choroid plexus of the lateral ventricles; it flows through the interventricular foramina to the third ventricle, along the Sylvian aqueduct (SA) to the fourth ventricle, proceeding to the subarachnoid space, and arriving in the spinal cord (Sartoretti et al., 2019; Liu et al., 2022; Vandenbulcke et al., 2022). Recently, the concept of the glymphatic pathway has been introduced (Iliff et al., 2012; Hablitz and Nedergaard, 2021). In the glymphatic system, CSF flows from the subarachnoid space to the periarterial space and enters the brain parenchyma, where it mixes with the interstitial fluid and exits into the paravenous space. CSF flow is pulsatile due to cardiac pulsations and breathing (Mestre et al., 2018; Liu et al., 2022). Depending on the cardiac cycle, CSF moves in opposite directions. During systole, intracranial arteries increase in volume which pushes CSF in the craniocaudal direction. The reverse happens in diastole as blood vessel volume decreases and CSF moves in the caudocranial direction (Lee et al., 2004; Vandenbulcke et al., 2022; Rohilla et al., 2023).

CSF serves several physiological functions. CSF provides support and cushioning for the brain. Movement of CSF through the brain parenchyma clears out waste and transports essential molecules (Howden et al., 2008; Nedergaard, 2013; Sartoretti et al., 2019; Liu et al., 2022; Vandenbulcke et al., 2022). These functions and dynamics, especially CSF flow and pressure, can be altered in pathologies. As part of the glymphatic system, CSF flow has been proposed to play a role in Alzheimer’s disease by failing to clear out amyloid beta and tau tangles, the characteristic protein markers (Stoquart-ElSankari et al., 2007; Liu et al., 2022; Reeves, 2025). Abnormal increase in intracranial pressure due to abnormal CSF drainage is associated with hydrocephalus (Lee et al., 2004; Stoquart-ElSankari et al., 2007; Rohilla et al., 2023; Reeves, 2025). Other studies have observed the roles of CSF flow in meningitis, cerebral edema, and other cerebrovascular diseases (Stoquart-ElSankari et al., 2007; Rohilla et al., 2023). Thus, measurement of CSF flow has been increasingly used for diagnostic and clinical monitoring purposes (Rohilla et al., 2023).

To explore the clinical implications of CSF flow, many studies focused on CSF flow through the SA which connects the third and fourth ventricles (Howden et al., 2008; Sartoretti et al., 2019). SA is easier to identify and image, so it has been studied frequently. On the other hand, the periarterial spaces of the circle of Willis (COW) and the middle cerebral artery (MCA), important conduits of CSF flow, remain unexplored. The COW is localized to the base of the brain and formed by anterior cerebral arteries, anterior communicating artery, internal carotid arteries (at their distal tips), posterior cerebral arteries, and posterior communicating artery. MCA originates from COW at the internal carotid arteries, joining and ascending in the lateral sulcus of the cerebrum, providing blood supply to many parts of the lateral cortex.

Therefore, this study has several objectives. First, because the CSF flow along the perivascular spaces of COW and MCA has not been examined, we aim to establish baseline measurements of CSF flow parameters along these conduits, including potential differences due to sex and age. Second, we aim to compare COW and MCA CSF flow parameters to the well-established flow through the SA.

To analyze CSF flow, we developed a new, semi-automated method for outlining perivascular spaces and extracting CSF flow parameters (Supplementary Figure 1). Many semi-automated or automated methods have been developed in order to increase accuracy, save time on analysis, and/or contribute to study reproducibility (Flórez et al., 2006; Yoshida et al., 2009). We conducted preliminary evaluation of our semi-automated program through validation with CSF flow parameters in the SA which are extensively covered in the literature.

We focused on the following CSF flow parameters: stroke volume (StroVol), volumetric flow rate (VFR), systolic flow rate (SFR), diastolic flow rate (DFR), velocity, peak systolic velocity (PSV), and peak diastolic velocity (PDV). While the most useful parameters have yet to be determined, we chose these parameters to capture the overall picture of CSF flow dynamics in the hopes that they may be used for future clinical diagnostics and therapeutics (Wagshul et al., 2006).

2 Materials and methods

2.1 Participants

The 24 healthy participants were recruited to achieve an even distribution by age and gender (13 males, 11 females). Ages ranged from 23 to 59 with a mean of 40 ± 11 (Table 1). To obtain baseline parameters by demographics, participants were separated into different age groups: 20–29, 30–39, 40–49, and 50–59. The study was approved by the Houston Methodist Hospital Institutional Review Board, protocol Pro00022145.

Table 1

Table 1. Participant demographics.

2.2 MRI acquisition

For this study, we used phase-contrast magnetic resonance imaging (PC-MRI). PC-MRI was first used in Barkhof et al. (1994) to evaluate the effect of age on SA flow. Since then, PC-MRI has emerged as the gold standard for measuring CSF flow (Liu et al., 2024). The advantages of PC-MRI are it is noninvasive and relatively quick at taking measurements (Stoquart-ElSankari et al., 2007).

A semi-automatic analysis method was developed to streamline region selection and ensure consistent CSF flow quantification across subjects. This tool integrates anatomical reference points and threshold-based segmentation to minimize operator bias. Future studies may extend this approach to other anatomical regions and validate its robustness further.

A 2D PC-MRI image was acquired at each region of interest (SA and the perivascular spaces of COW and MCA) chosen by an experienced neuroscientist (EG, ARG). Details of the acquisition parameters are as follows: slice thickness 3 mm, acquisition matrix 272 × 272, FOV 160 cm 160 cm, in-plane resolution 0.6 mm x 0.6 mm. TE = 3 msec, TR = 105 msec. Images were acquired with a single transmit, 32 channel receive head coil in the FDA-approved clinical mode of the MAGNETOM 7 Tesla human MRI scanner (Siemens Healthineers, Erlangen, Germany). A 3D velocity encoding scheme was used to account for CSF flow direction not exactly parallel to the slice normal. The encoding velocity (VENC) was set at 5 cm/s to remain sensitive to the corresponding flow velocity of CSF flow and to eliminate blood flow contamination, recognized as global maximum/minimum gray scale values, due to the phase wrapping artifact of blood velocities exceeding this VENC value. A total of 20 images evenly spaced across the cardiac cycle were reconstructed with a total acquisition time of approximately 5 min for each location.

2.3 CSF flow analysis

As the VENC value was low compared to blood flow velocities, global maximum black/white gray scale intensity values—caused by the phase wrapping artifact—were used as a mask for distinguishing and eliminating blood flow from perivascular CSF flow in 2D phase contrast images. Regions of interest (ROIs) in an annular shape centered on the arterial cross section were chosen for quantifying CSF flow using a semi-automated algorithm described in Supplementary Figure 1.

For the first phase image of the 20 series of images for the cardiac cycle, the X, Y-coordinates, tolerance, and band size of the ROIs were determined by the user (Supplementary Figure 1).

Magnitude images were used as anatomical references. Two ROIs—one left and one right—were identified for each COW and MCA, and one ROI was identified for SA (Figure 1). The established ROI settings were entered into our developed in-house semi-automated program to produce flow rate (FR) and Velocity values. FR (mL/s) was calculated as the $\text{Area of the}\mathrm{ROI}\left(\text{converted from}m{m}^2\text{to}c{m}^2\right)\times \text{VENC}\times \left(\frac{\text{mean intensity}}{\text{maximum gray scale intensity}}\right).$ Velocity (cm/s) was calculated as $\mathrm{FR}÷\text{area of the}\mathrm{ROI}$ . ROIs for 20 cardiac gated images over one cardiac cycle were obtained from each series to obtain the FR and Velocity curves.

Figure 1

Figure 1. Examples of ROIs. (A) ROI of SA, (B) one of two ROIs of COW, (C) and one of two ROIs of MCA on phase images.

Peak timing was standardized by looking at which percentage of the cardiac cycle peak flow occurred. The percentage of the cardiac cycle was calculated as $\left(\frac{\text{the ordinal position of the image within the series}}{20}\right)\times 100$ . The stroke volume was calculated as $\mathrm{FR}\times \text{time}$ .

2.4 Statistical analysis

For all parameters in each conduit of SA, COW, and MCA, median values and bias-corrected and accelerated bootstrap confidence intervals (CIs) from non-parametric bootstrapping were calculated.

To compare the conduits and evaluate the influence of demographics, various models were constructed. For StroVol, VFR, and Velocity, linear mixed models (LMM) were constructed with block bootstrapping. Generalized linear mixed models (GLMM) with the gamma family and log link function were constructed with block bootstrapping for SFR, DFR, and Area. Generalized linear models (GLM) were constructed for PSV and PDV. For these LMMs, GLMMs, and GLMs, accelerated bootstrap CIs were calculated. For time to peak as percentage of cardiac cycle duration (TP) of StroVol (TPStroVol), VFR (TPVFR), and Velocity (TPVelocity), beta regression with the logit link function was constructed, and Wald CIs were calculated. For comparing SA, COW, and MCA, fixed effects were the intercept (= SA), time (except for PSV, PDV, and TP parameters), COW, and MCA. SA was set as the intercept because CSF flow parameters for the SA are already established in the literature. For evaluating the influence of sex and age, fixed effects were the intercept, time (except for PSV, PDV, and TP parameters), sex, and age. An interaction term between sex and age was also added. When applicable, participants were considered as random effects to account for the non-independence of time series. Age was rescaled, zero random effects were dropped, and/or optimizers were changed to address model convergence issues as needed. Whenever the data contained zero-values, the following transformation was applied to the outcome variable before beta regression: $\frac{(\text{outcome variable}\times 23+0.5)}{23}$ . For post-hoc testing, p-values were adjusted using the Benjamini-Hochberg procedure. Assumptions were checked with QQ plots and residual plots. All statistical analysis and visualization were performed in R/RStudio.

3 Results

Baseline values of CSF flow parameters with the median and 95% CI by conduits and by sex and age are reported in Tables 2–4, respectively. Model estimates of the fixed and random effects are reported in Table 5 for conduits comparison and Supplementary Table 1 for the effect of sex and age. StroVol is reported in mL. VFR, SFR, and DFR are reported in mL/s. Velocity, PSV, and PDV are reported in cm/s. TPStroVol, TPVFR, and TPVelocity are reported in percentage of the cardiac cycle. Area is reported in cm².

Table 2

Table 2. Median values of CSF flow parameters in SA, COW, and MCA.

Table 3

Table 3. Median values of CSF flow parameters in different conduits by sex.

Table 4

Table 4. Median values of CSF flow parameters in different conduits by age group.

Table 5

Table 5. Estimates of models of CSF flow parameters in SA (intercept) vs. COW vs. MCA.

As expected, VFR oscillates in the SA which reflects cardiac pulsations. On the other hand, COW VFR decreases throughout the cardiac cycle (Figure 2A). Interestingly, Velocity in the SA is pulsatile while it stays relatively constant in the COW and MCA (Figure 2B).

Figure 2

Figure 2. Aggregate CSF flow curves. Aggregation of CSF (A) VFR and (B) velocity curves for all participants.

SA was set as the intercept because CSF flow parameters for this conduit are already established in the literature. The StroVol LMM indicates strong effects with COW StroVol 0.0150 mL (95% CI [0.0103, 0.0190], p = 0.00400) and MCA StroVol 0.0142 mL (95% CI [0.00979, 0.0194], p = 0.00400) higher than SA (Table 5). The TPStroVol occurs later in the cardiac cycle in COW (β = 1.12, 95% CI [0.792, 1.45], p = 7.34 × 10–11) and MCA (β = 0.966, 95% CI [0.646, 1.29], p = 1.19 × 10–8) compared to SA (Table 5).

The LMMs and GLMMs for flow also indicate strong effects of the conduit as the magnitude of the parameters is higher in COW and MCA. Specifically, COW VFR is 0.0350 mL/s (95% CI [0.0261, 0.0460], p = 0.00400) and MCA VFR is 0.0310 mL/s (95% CI [0.0203, 0.0411], p = 0.00400) higher than SA (Table 5). COW SFR increases by a factor of 1.36 (95% CI [1.20, 1.50], p = 0.00400) and MCA SFR by a factor of 1.36 (95% CI [1.21, 1.51], p = 0.00400) compared to SA (Table 5). COW DFR also increases by a factor of 1.48 (95% CI [1.16, 1.83], p = 0.00799) and MCA SFR by a factor of 1.61 (95% CI [1.33, 1.94], p = 0.00799) compared to SA (Table 5). The TPVFR also occurs later in COW (β = 0.690, 95% CI [0.322, 1.06], p = 0.000468) and MCA (β = 0.486, 95% CI [0.110, 0.862], p = 0.0274) compared to SA (Table 5).

Velocity in the LMM and GLMs also exhibits strong effects, and the magnitude of the parameters is lower in COW and MCA. COW Velocity is 0.159 cm/s (95% CI [−0.286, −0.0498], p = 0.0160) and MCA Velocity is 0.198 cm/s (95% CI [0.0204, 0.0410], p = 0.00599) lower than SA (Table 5). COW PSV decreases by a factor of 0.773 (95% CI [−0.954, −0.613], p = 0.00400), and MCA PSV decreases by a factor of 0.955 (95% CI [−1.12, −0.773], p = 0.00400) compared to SA (Table 5). COW PDV also decreases by a factor of 0.984 (95% CI [−1.24, −0.721], p = 0.00400), and MCA PDV decreases by a factor of 1.01 (95% CI [−1.29, −0.747], p = 0.00400) compared to SA (Table 5). For TPVelocity, we were unable to find evidence against the hypothesis that SA = COW = MCA (p = 0.861) (COW β = 0.0137, 95% CI [−0.286, 0.313]) (MCA β = 0.0846, 95% CI [−0.224, 0.393]) (Table 5).

For most CSF flow parameters, COW and MCA values are similar (Table 5). It should be noted that the 95% CI and p-value (p = 0.0280) for PSV contradict each other, but the latter is more directly related to our hypothesis testing (Table 5). Therefore, we found evidence against the hypothesis that COW = MCA for PSV.

There is considerably greater intra-subject variation than between-subject variation for the LMMs and GLMMs which is unsurprising given the pulsatile nature of CSF flow (Table 5). These findings further support our observations in healthy individuals as the effects are comparable.

3.1 Sex and age are weak predictors of CSF flow parameters

For all CSF flow parameters, sex, age, and their interaction are on the orders of magnitude less than the intercept (Supplementary Table 1). Moreover, we did not find evidence against the hypothesis that Sex = 0, Age = 0, and their interaction = 0 (Supplementary Table 1). While the 95% CIs and p-values of SA DFR (p = 0.996), COW DFR (p = 0.548), MCA DFR (p = 0.999), and MCA TPStroVol (p = 0.108) contradict each other, we will defer to the latter as explained previously (Supplementary Table 1). Therefore, we can say with great certainty that the effects of sex, age, and their interaction are weak.

4 Discussion

Our results compare favorably with previously reported CSF flow characteristics. We demonstrated pulsatile CSF flow and SA VFR (0.00700 mL/s) in the range of literature values, 0.0049–0.0432 mL/s (Table 2 and Figure 2) (Lee et al., 2004; Wagshul et al., 2006; Yoshida et al., 2009; Oner et al., 2017). However, Flórez et al. (2006) found SA VFR was 0.0635 mL/s in healthy participants. This difference may be attributed to Flórez et al. (2006) use of background correction and their own semi-automated program for creating ROIs. SA Velocity also fell into the reported range of Ståhlberg et al. (1989) study: 0–3 cm/s (Table 2). Overall, these results support the use of our semi-automated program.

SA peak velocities were either similar or lower compared to the literature. SA PSV (1.45 cm/s) and the magnitude of SA PDV (0.867 cm/s) ranged from 2.0 to 11.5 cm/s in the literature (Table 2) (Ståhlberg et al., 1989; Lee et al., 2004; Flórez et al., 2006; Tulupov et al., 2011). Differences may be attributed to the use of tolerance in our semi-automated program and commonly known partial volume effects which cause reduction in maximum flow values, both contributing to underestimated peak velocities (Supplementary Figure 1).

Generally, differences in CSF flow parameters may also be due to differences in MRI manufacturer, artifacts, resolution levels, and VENC (Vandenbulcke et al., 2022). Although CSF velocities observed in our study are low, previous literature has demonstrated that careful selection of low VENC values (e.g., 5 cm/s) in 2D PC-MRI allows for reliable quantification of slow perivascular flow. We acknowledge that novel methods such as IVIM or 4D flow MRI may offer additional insight into global CSF dynamics, but these methods come at the cost of longer scan times and increased complexity. Our goal was to achieve high measurement reliability in targeted anatomical locations which justifies our methodological choice.

While 4D flow MRI provides a comprehensive 3D mapping of CSF dynamics, it is associated with longer acquisition times and increased post-processing demands; in contrast, 2D PC-MRI is a well-validated and reliable method that allows precise and reproducible quantification of CSF velocities in specific anatomical regions of interest. Given the aim of this study is to establish normative reference values in well-defined perivascular regions, 2D PC-MRI was considered the most suitable technique.

What is unique in this study is the thorough examination of CSF flow in the perivascular spaces of COW and MCA. Since our semi-automated program was supported by the SA results, we can now expand its use to measurements of the COW and MCA. For all CSF flow parameters, there is ample evidence they are different in the perivascular spaces of COW and MCA than in the SA (Table 5). Since the magnitude of flow parameters in the COW and MCA is greater while the magnitude of velocity parameters is lower than in the SA, it is plausible the size of these conduits is driving the increased flow rates in the COW and MCA compared to SA (Table 5). The cross-sectional area of the COW and MCA perivascular spaces is larger than SA, so this finding is unsurprising. Between the COW and MCA however, there was minimal difference for most parameters (Table 5). Since fluid flows through the COW and enters the MCA shortly after, this result was expected.

Besides conduits, demographics may also influence CSF flow dynamics. We used our semi-automated program to establish baseline values of CSF flow parameters for each sex and age group. Moreover, we looked at sex, age, and their interaction as predictors. Across the board though, these effects were negligible. Thus, sex and/or age seems to have minimal influence on CSF flow dynamics. Like our study, Sartoretti et al. (2019) used regression models with sex and age as predictors, and they found sex and age could only explain a small part of CSF flow parameters which they quantified to be 6–18%. Furthermore, they found sex and age were not significant predictors for the SA Velocity. Other studies found similar results where sex and age were not significant predictors for the SA VFR, SFR, DFR, Velocity, and Peak Velocity (Flórez et al., 2006; Unal et al., 2009; Oner et al., 2017; Hett et al., 2022).

Some studies, however, have found sex and age dependencies of several CSF flow parameters. Sartoretti et al. (2019) found these predictors were significant for the SA VFR as well as SA peak velocity. The SA VFR and peak velocity increased with age and was higher in males (Sartoretti et al., 2019). Unal et al. (2009) also observed the age dependence of SA peak velocity, but the relationship was inverse. Stoquart-ElSankari et al. (2007) found the age dependence for SA VFR, but similarly, the trend was downward. Rohilla et al. (2023) found weak positive linear correlations with age for the SA SFR and PDV and moderate positive linear correlations for the SA DFR and PSV. This variation in results may be due to the age range of participants and how they were divided into groups. Rohilla et al. (2023) study had participants from 40 to 78 years of age while our study had participants ranging from 23 to 59 years of age. The elderly group in Stoquart-ElSankari et al. (2007) study had a mean age of 71 while the young group had a mean age of 27.5. The difference in CSF flow parameters between these groups may be more obvious because of the higher prevalence of chronic conditions among the elderly. Our study, though, only looked at healthy, relatively young participants.

Our results may have also differed because of our limited sample size. However, Sartoretti et al. (2019) comprehensive study had 128 healthy participants from 17 to 88 years of age which found similar results (Sartoretti et al., 2019). As both of our studies suggest, other factors may influence CSF flow dynamics to a greater extent, including cardiac pulsations, breathing, anatomy of brain, and size of blood vessels (Lee et al., 2004; Sartoretti et al., 2019).

Limitations of our study include eddy currents and partial volume effects. Thus, MRI protocols should be optimized, and the effect of different VENCs on CSF flow parameter measurements should be evaluated. Another major limitation is the inaccuracy of the ROI delineation process. Our semi-automated program used for that process could potentially introduce variability in measurements (Supplementary Figure 1). Furthermore, it is only capable of capturing one continuous ROI. Thus, if conduits appear in multiple areas of the MRI phase image as in the case of the COW and MCA, measurements would be underestimated. To improve the ROI delineation process, our semi-automated program should be formally evaluated, and its inter-observer reliability should be measured (Supplementary Figure 1). We believe these efforts would be beneficial as automation has the benefits of increasing accuracy, reproducibility, and the ease of studying large samples (Hett et al., 2022). Lastly, it is clear there is no consensus on the effects of sex, age, and their interaction, even in literature on the SA. Further studies, then, need to be conducted and particularly focus on comparing the elderly population to younger populations.

In this study, we have established for the first time the baseline values for the perivascular spaces of COW and MCA CSF flow parameters and compared them to those in the SA. We also highlighted the limited influence of sex and/or age. Future studies can use this research as a starting point to investigate the CSF flow in the perivascular spaces of COW and MCA, thereby increasing the accuracy of parameter measurements. It might also be helpful to look at sex, age, and other factors (e.g., breathing) simultaneously to get a better sense of what drives CSF flow dynamics. Finally, these studies should be repeated in patients with central nervous system or cerebrovascular system pathologies which could potentially lead to the applications of CSF flow to clinical diagnosis, monitoring, and treatment.

Data availability statement

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding authors.

Ethics statement

The studies involving humans were approved by Houston Methodist Hospital Institutional Review Board. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

Author contributions

RL: Visualization, Investigation, Resources, Writing – review & editing, Formal analysis, Writing – original draft, Data curation, Software, Methodology. CK: Software, Writing – review & editing, Methodology, Writing – original draft, Data curation, Visualization, Resources, Formal analysis. AR-G: Methodology, Writing – original draft, Supervision, Data curation, Conceptualization, Writing – review & editing, Project administration, Investigation. EG: Data curation, Methodology, Writing – original draft, Supervision, Investigation, Resources, Conceptualization, Project administration, Writing – review & editing. GB: Project administration, Writing – original draft, Supervision, Conceptualization, Writing – review & editing, Resources, Funding acquisition.

Funding

The author(s) declare that financial support was received for the research and/or publication of this article. This study was partially funded by the Siemens Healthineers. The funder was not involved in the study design, collection, analysis, interpretation of data, the writing of this article, or the decision to submit it for publication.

Acknowledgments

We would like to thank Dr. Ioannis Konstantinidis, University of Houston, for his guidance on statistical analysis. Preprint appeared in bioRxiv (Le et al., 2025).

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declare that no Gen AI was used in the creation of this manuscript.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnins.2025.1622351/full#supplementary-material

References

Barkhof, F., Kouwenhoven, M., Scheltens, P., Sprenger, M., Algra, P., and Valk, J. (1994). Phase-contrast cine mr imaging of normal aqueductal CSF flow: effect of aging and relation to CSF void on modulus MR. Acta Radiol. 35, 123–130. doi: 10.1177/028418519403500204

PubMed Abstract | Crossref Full Text | Google Scholar

Flórez, Y. N., Moratal, D., Forner, J., Martí-Bonmatí, L., Arana, E., Guajardo-Hernández, U., et al. (2006). Semiautomatic analysis of phase contrast magnetic resonance imaging of cerebrospinal fluid flow through the aqueduct of Sylvius. Magn Reson Mater Phy 19, 78–87. doi: 10.1007/s10334-006-0030-6

PubMed Abstract | Crossref Full Text | Google Scholar

Hablitz, L. M., and Nedergaard, M. (2021). The Glymphatic system: a novel component of fundamental neurobiology. J. Neurosci. 41, 7698–7711. doi: 10.1523/JNEUROSCI.0619-21.2021

PubMed Abstract | Crossref Full Text | Google Scholar

Hett, K., McKnight, C. D., Eisma, J. J., Elenberger, J., Lindsey, J. S., Considine, C. M., et al. (2022). Parasagittal dural space and cerebrospinal fluid (CSF) flow across the lifespan in healthy adults. Fluids Barriers CNS 19:24. doi: 10.1186/s12987-022-00320-4

PubMed Abstract | Crossref Full Text | Google Scholar

Howden, L., Giddings, D., Power, H., Aroussi, A., Vloeberghs, M., Garnett, M., et al. (2008). Three-dimensional cerebrospinal fluid flow within the human ventricular system. Comput. Methods Biomech. Biomed. Engin. 11, 123–133. doi: 10.1080/10255840701492118

PubMed Abstract | Crossref Full Text | Google Scholar

Iliff, J. J., Wang, M., Liao, Y., Plogg, B. A., Peng, W., Gundersen, G. A., et al. (2012). A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid β. Sci. Transl. Med. 4:748. doi: 10.1126/scitranslmed.3003748

PubMed Abstract | Crossref Full Text | Google Scholar

Le, R. F. D., Karmonik, C., Regnier-Golanov, A. S., Golanov, E. V., and Britz, G. W. (2025). Quantitative evaluation of normal cerebrospinal fluid flow in the Sylvian aqueduct and perivascular spaces of the middle cerebral artery and circle of Willis via 2D phase-contrast MR imaging. bioRxiv.

Google Scholar

Lee, J. H., Lee, H. K., Kim, J. K., Kim, H. J., Park, J. K., and Choi, C. G. (2004). CSF flow quantification of the cerebral aqueduct in Normal volunteers using phase contrast cine MR imaging. Korean J. Radiol. 5, 81–86. doi: 10.3348/kjr.2004.5.2.81

PubMed Abstract | Crossref Full Text | Google Scholar

Liu, G., Ladrón-de-Guevara, A., Izhiman, Y., Nedergaard, M., and Du, T. (2022). Measurements of cerebrospinal fluid production: a review of the limitations and advantages of current methodologies. Fluids Barriers CNS 19:101. doi: 10.1186/s12987-022-00382-4

PubMed Abstract | Crossref Full Text | Google Scholar

Liu, P., Owashi, K., Monnier, H., Metanbou, S., Capel, C., and Balédent, O. (2024). Validating the accuracy of real-time phase-contrast MRI and quantifying the effects of free breathing on cerebrospinal fluid dynamics. Fluids Barriers CNS 21:25. doi: 10.1186/s12987-024-00520-0

PubMed Abstract | Crossref Full Text | Google Scholar

Mestre, H., Tithof, J., Du, T., Song, W., Peng, W., Sweeney, A. M., et al. (2018). Flow of cerebrospinal fluid is driven by arterial pulsations and is reduced in hypertension. Nat. Commun. 9:4878. doi: 10.1038/s41467-018-07318-3

PubMed Abstract | Crossref Full Text | Google Scholar

Nedergaard, M. (2013). Garbage truck of the brain. Science 340, 1529–1530. doi: 10.1126/science.1240514

PubMed Abstract | Crossref Full Text | Google Scholar

Oner, Z., Sagіr Kahraman, A., Kose, E., Oner, S., Kavaklі, A., Cay, M., et al. (2017). Quantitative evaluation of Normal Aqueductal cerebrospinal fluid flow using phase-contrast cine MRI according to age and sex. Anat. Rec. 300, 549–555. doi: 10.1002/ar.23514

PubMed Abstract | Crossref Full Text | Google Scholar

Reeves, B. C. (2025). Glymphatic system impairment in Alzheimer’s disease and idiopathic Normal pressure hydrocephalus. Trends Mol. Med. 26, 285–295.

Google Scholar

Rohilla, S., Kumar, P., and Singh, I. (2023). Cerebrospinal fluid flow parameters in normal subjects above 40 years of age. Indian J. Radiol. Imaging 34:413. doi: 10.1055/s-0043-1776413

PubMed Abstract | Crossref Full Text | Google Scholar

Sartoretti, T., Wyss, M., Sartoretti, E., Reischauer, C., Hainc, N., Graf, N., et al. (2019). Sex and age dependencies of Aqueductal cerebrospinal fluid dynamics parameters in healthy subjects. Front. Aging Neurosci. 11:199. doi: 10.3389/fnagi.2019.00199

PubMed Abstract | Crossref Full Text | Google Scholar

Ståhlberg, F., Mogelvang, J., Thomsen, C., Nordell, B., Stubgaard, M., Ericsson, A., et al. (1989). A method for MR quantification of flow velocities in blood and CSF using interleaved gradient-echo pulse sequences. Magn. Reson. Imaging 7, 655–667. doi: 10.1016/0730-725X(89)90535-3

PubMed Abstract | Crossref Full Text | Google Scholar

Stoquart-ElSankari, S., Balédent, O., Gondry-Jouet, C., Makki, M., Godefroy, O., and Meyer, M.-E. (2007). Aging effects on cerebral blood and cerebrospinal fluid flows. J. Cereb. Blood Flow Metab. 27, 1563–1572. doi: 10.1038/sj.jcbfm.9600462

PubMed Abstract | Crossref Full Text | Google Scholar

Tulupov, A. A., Bogomyakova, O. B., Savelyeva, L. A., and Prygova, Y. A. (2011). Quantification of flow of cerebrospinal fluid on basal level of brain by a phase-contrast MRI technique. Appl. Magn. Reson. 41, 543–550. doi: 10.1007/s00723-011-0291-8

Crossref Full Text | Google Scholar

Unal, O., Kartum, A., Avcu, S., Etlik, O., Arslan, H., and Bora, A. (2009). Cine phase-contrast MRI evaluation of normal aqueductal csf flow according to sex and age. Diagn. Interv. Radiol. 15:4. doi: 10.4261/1305-3825.DIR.2321-08.1

PubMed Abstract | Crossref Full Text | Google Scholar

Vandenbulcke, S., De Pauw, T., Dewaele, F., Degroote, J., and Segers, P. (2022). Computational fluid dynamics model to predict the dynamical behavior of the cerebrospinal fluid through implementation of physiological boundary conditions. Front. Bioeng. Biotechnol. 10:1040517. doi: 10.3389/fbioe.2022.1040517

PubMed Abstract | Crossref Full Text | Google Scholar

Wagshul, M. E., Chen, J. J., Egnor, M. R., McCormack, E. J., and Roche, P. E. (2006). Amplitude and phase of cerebrospinal fluid pulsations: experimental studies and review of the literature. JNS 104, 810–819. doi: 10.3171/jns.2006.104.5.810

PubMed Abstract | Crossref Full Text | Google Scholar

Yoshida, K., Takahashi, H., Saijo, M., Ueguchi, T., Tanaka, H., Fujita, N., et al. (2009). Phase-contrast MR studies of CSF flow rate in the cerebral aqueduct and cervical subarachnoid space with correlation-based segmentation. MRMS 8, 91–100. doi: 10.2463/mrms.8.91

PubMed Abstract | Crossref Full Text | Google Scholar