Region specific drivers of cerebrospinal fluid mobility as measured by high-resolution non-invasive MRI in humans

Region specific drivers of cerebrospinal fluid mobility as measured by high-resolution non-invasive MRI in humans | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Region specific drivers of cerebrospinal fluid mobility as measured by high-resolution non-invasive MRI in humans Lydiane Hirschler, Bobby A. Runderkamp, Andreas Decker, Thijs W. van Harten, and 19 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3178346/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 14 Oct, 2025 Read the published version in Nature Neuroscience → Version 1 posted You are reading this latest preprint version Abstract Highlights CSF-mobility can be measured in humans down to the level of perivascular spaces surrounding penetrating vessels using CSF-STREAM In perivascular spaces surrounding penetrating vessels, the cardiac and respiratory cycles have similar effects on CSF-mobility Entraining vasomotion at 0.1 Hz can drive CSF-mobility in humans CSF-mobility is altered in patients with cerebral amyloid angiopathy Abstract Many neurological diseases are characterized by the accumulation of toxic proteins in the brain. This accumulation has been associated with improper clearance from the parenchyma. Recent discoveries highlighted perivascular spaces, which are cerebrospinal fluid (CSF)-filled spaces, as the channels of brain clearance. The forces driving CSF-mobility within perivascular spaces are still debated. Here, we present a new, non-invasive, CSF-specific magnetic resonance imaging technique (CSF-STREAM), that enables detailed in vivo measurement of CSF-mobility in humans, for the first time down to the level of perivascular spaces located around penetrating vessels, i.e. close to protein production sites. We find region-specific drivers for CSF-mobility and demonstrate that CSF-mobility can be increased by entraining vasomotion. Furthermore, we found region-specific CSF-mobility alterations in patients with cerebral amyloid angiopathy, a brain disorder associated with clearance impairment. The availability of this new technique opens up avenues to investigate the impact of CSF-mediated clearance in neurodegeneration and sleep. Biological sciences/Biological techniques/Imaging/Magnetic resonance imaging Biological sciences/Physiology/Neurophysiology Biological sciences/Neuroscience/Neuro–vascular interactions Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Highlights CSF-mobility can be measured in humans down to the level of perivascular spaces surrounding penetrating vessels using CSF-STREAM In perivascular spaces surrounding penetrating vessels, the cardiac and respiratory cycles have similar effects on CSF-mobility Entraining vasomotion at 0.1 Hz can drive CSF-mobility in humans CSF-mobility is altered in patients with cerebral amyloid angiopathy Introduction Due to its high metabolic rate, the brain produces large quantities of proteins, whose abnormal accumulation is involved in a number of pathologies and neurodegenerative disorders. However, unlike other organs in the body, the brain tissue lacks a classic lymphatic system to transport excess soluble proteins it produces (‘waste’) out of the brain (e.g. to lymph nodes or dural lymphatic vessels). The true nature of brain clearance mechanisms has eluded characterization for centuries, until the recent uptick in interest in the topic 1,2 . The use of microscopy in rodents 3–5 and invasive intrathecal injections in humans 6,7 has allowed researchers to unravel and describe many aspects of new brain clearance pathways and more subtle physiological processes, and their results have opened up new questions and debate. While we know that brain clearance processes must involve cerebrospinal fluid (CSF) as the main carrier of ‘waste‘ products, likely along pathways surrounding small blood vessels called perivascular spaces (PVS) 8 , the exact anatomical pathway(s), driving force(s) as well as physiological processes involved in CSF-mediated brain clearance remain unresolved 1,9,10 . Some studies suggest the presence of an active mechanism driving CSF-flow along PVS (including the glymphatic 3 and intramural periarterial drainage 11 theories), whereas others propose that perivascular clearance mainly occurs via more passive mixing mechanisms 9,12 . In all proposed mechanisms, CSF-mobility in PVS would be facilitated by physiological motion processes, such as cardiac pulsations 4 , respiration 13 , or vasomotion 5 . As such, it has been suggested that these drivers of motion can propel soluble ‘waste’ products from the PVS up towards the pial surface, where bulk flow assures further egress. Studying CSF-mediated brain clearance and its driving forces is of particular importance since clearance failure has been implicated in the accumulation of toxic proteins in the brain, such as amyloid-β and tau. A plethora of neurological diseases like Alzheimer’s disease 14 , cerebral amyloid angiopathy (CAA) 15 , traumatic brain injury 16 , and ischemic stroke 17 are associated with brain clearance deficiencies. CAA, especially, is a common small vessel disease and leading cause of hemorrhagic stroke and dementia in older individuals, characterized by the accumulation of amyloid-β in the vessel wall, possibly due to impaired CSF-mediated amyloid-β clearance 15,18 . CAA frequently co-occurs with Alzheimer’s pathology and is associated with increased risk of developing amyloid related imaging abnormalities in the context of anti-amyloid immunotherapy 15 . Notably, in a rat model of CAA, the mobility of CSF in the subarachnoid space (SAS) surrounding large arteries was recently found to be increased 19 . This was accompanied by a reduction of volume of tracer transport to the brain tissue, altogether suggesting that CSF would bypass the brain tissue due to amyloid-β deposits. However, it is currently unknown whether these recent findings in rodents also translate to humans with CAA. A better understanding of CSF-mediated brain clearance and its driving forces is therefore urgent, as it would provide crucial new avenues not only to elucidate the pathophysiology of these neurological diseases, but also towards new targets for efficient therapeutic strategies to slow or stop disease progression. Unfortunately, current knowledge of CSF-mediated brain clearance mechanisms is mostly derived from experimental studies performed in rodents, associated with significant limitations. First, these studies tend to use techniques which introduce perturbations to the very system they aim to characterize, such as euthanasia prior to measurement, which may result in the collapse of essential structures for brain clearance 4 , anesthesia, which interferes with hemodynamics and likely with brain clearance as well, or invasive imaging techniques 3–5,20 like cranial windows and injection of fluorescent dyes, which may induce local pressure changes and thereby affect physiological CSF motion. Second, it remains unclear how experimental findings in rodents translate to humans given the inherent differences in brain size and physiological parameters between species 2 . For example, cardiac frequency is about seven to twelve times slower in humans compared to mice 21 , whereas vasomotion frequency is comparable, centered around 0.1 Hz. This could influence the relative contributions to the suggested driving forces of CSF flow, i.e. the cardiac cycle, respiration, and vasomotion. Moreover, to the best of our knowledge, current in vivo studies only focus on motion of CSF in the SAS and in the ventricular system 22–26 , lacking the spatial resolution to investigate CSF-mobility in PVS around penetrating vessels, which are believed to be the channels along which soluble ‘waste’ products clear out of the brain. A non-invasive method that measures CSF-mobility at a high spatial resolution is needed to further understand perivascular brain clearance mechanisms in humans, allowing the assessment of CSF-mobility directly where clearance is thought to occur, i.e. within CSF-filled PVS. This would pave the way for studying CSF-mediated brain clearance in larger sample sizes, patient cohorts, and in longitudinal follow-up studies. Magnetic resonance imaging (MRI) at a high magnetic field strength (7 Tesla) is an excellent modality for such a non-invasive imaging strategy, as it provides the necessary resolution to image the PVS and has the added benefit of easily differentiating CSF from other brain tissues or fluids by exploiting its specific magnetic properties. In this study, we present a non-invasive, high-resolution, and CSF-specific MRI technique that allows the characterization of CSF-mobility, even in PVS around penetrating arteries. We also investigate and compare the influence of the cardiac cycle, respiratory cycle, and vasomotion as driving forces for CSF-mobility. Lastly, we apply the newly developed technique in patients with CAA in a pilot study, in order to explore the potential of this technique to improve our understanding of neurodegenerative diseases. Results High-resolution imaging of CSF-mobility is achieved using CSF-STREAM Whole-brain CSF-signal was visualized in twenty healthy, younger individuals (age 33±12 years, 16 females, 4 males) at rest using ultra-high field (7 Tesla) MRI with a T 2 -prepared high-resolution (i.e. 0.45 mm isotropic voxel-size) accelerated readout with a long echo-time (Fig. 1A-B, Fig. 2A, Extended data Fig. 1 and Suppl. video 1). Importantly, signal originating from blood and brain tissue was suppressed (i.e. not significantly different from the noise level), such that the CSF-signal was successfully isolated (Extended data Fig. 1). The introduction of motion-sensitizing gradients of 3.5 mm/s in the T 2 -preparation module, which during repeated measurements encode mobility in six orthogonal directions, allowed calculation of a tensor from which the CSF-mobility, fractional anisotropy (FA) and principal CSF-mobility orientation were computed (Fig. 1). CSF-mobility is measured and calculated in a similar way as an apparent diffusion coefficient acquired at a very low b-value, making the MRI sequence more sensitive to flow than diffusion 27 . The term ‘mobility’ is used as opposed to ‘flow’ or ‘diffusion’ to accentuate that the dephasing underlying the signal attenuation is caused by either slow flow, laminar flow or by back-and-forth motion of CSF, or a combination of all processes 22,27 . Altogether, the proposed CSF-Selective T 2 -prepared REadout with Acceleration and Mobility-encoding (CSF-STREAM) provides a fully non-invasive technique to quantitatively measure the mobility of CSF at an unprecedented high resolution: from the ventricles to the SAS around large vessels, down to small PVS in the basal ganglia and around penetrating arteries (Fig. 2 and Extended data Fig. 2). In the SAS around the MCA and in PVS, CSF mainly moves along the vessel as represented by the principal vector orientation (Fig. 1C-D). The FA in PVS (Fig. 2B) was high, further substantiating that CSF preferentially moves along one orientation. CSF-mobility was higher at the base of the brain than at the brain surface (Fig. 1E, Fig. 2B). It was highest in the SAS around the middle cerebral artery (0.041±0.008 mm 2 /s) and three times lower in PVS of the basal ganglia and of penetrating arteries (0.012±0.003 mm 2 /s and 0.015±0.005 mm 2 /s, respectively). Region- specific effects of cardiac and respiratory cycles on CSF-mobility Next, the influence of driving forces of CSF-mobility was assessed in eleven individuals by studying the effect of the cardiac and respiratory cycles on CSF-mobility, as these are proposed to be driving forces of CSF-mobility 2 . After retrospective binning of k-space to the recordings of external physiological sensors (Extended data Fig. 3), we observed that CSF-mobility and FA were indeed fluctuating across the cardiac and respiratory cycles in an oscillatory manner (Fig. 3, Fig. 4). We found that in selected large CSF-spaces located at the base of the brain, i.e. the SAS around the MCA (SAS-MCA) and the 4 th ventricle, the cardiac cycle was associated with significantly larger CSF-mobility oscillations than respiration. This is shown by the average CSF-mobility changes across phases in Fig. 5 and the significantly better fit quality to a sinusoid (SAS-MCA: 0.7±0.06 for cardiac versus 0.5±0.08 for respiration, p=0.002; 4 th ventricle: 0.8±0.08 for cardiac versus 0.6±0.1 for respiration, p=0.002; Wilcoxon signed rank tests) and CSF-mobility change amplitude (SAS-MCA: 3.2±0.6 % for cardiac versus 1.2±0.4 % for respiration, p<0.001; 4 th ventricle: 8.4±2.9 % for cardiac versus 2.5±1.0 % for respiration, p<0.001; Wilcoxon signed rank tests), as shown in Fig. 6. In contrast, in the smaller PVS in the basal ganglia and around penetrating arteries, CSF-mobility was driven by similar contributions of the cardiac and respiratory cycles (Fig. 5 and Fig. 6): the cardiac cycle induced a CSF-mobility change amplitude of 2.5±0.5 % in basal ganglia PVS and 2.8±0.7 % in PVS around penetrating arteries, whereas respiration led to 2.4±0.7 % (basal ganglia PVS) and 2.6±0.6 % (penetrating PVS) CSF-mobility change amplitudes. Our method of retrospective binning did not artificially generate the observed CSF-mobility changes, as random binning did not result in a clear oscillatory pattern across phases (Fig. 5): CSF-mobility changes across random phases in a zig-zag-like fashion with e.g. several dips and peaks within one cycle, indicating a noisier pattern than that observed across cardiac and respiration phases. This can also be concluded quantitatively from the fit quality, fit amplitude, and the amount of voxels with coherent changes across phases (i.e. voxels with R 2 >0.5), which were all significantly lower for random ordering compared to binning based on cardiac or respiratory traces (Fig. 6, Extended data Fig. 5). Moreover, spatially coherent patterns (e.g. a right-left symmetry) in CSF-mobility changes can be observed for cardiac and respiration binning but not for random ordering (Fig. 3 and Suppl. video 2), providing further confidence that these oscillations are indeed originating from physiological pulsations and are not an artificial result of the reconstruction step. Along with the cardiac and respiratory cycles, vasomotion has been proposed as one of the driving forces for CSF. Unfortunately, there is no external physiological sensor to detect vasomotion that could be used for retrospective binning. A previous study in rodents showed that a 0.1 Hz visual stimulation could entrain vasomotion and lead to faster clearance of fluorescent tracers in the mouse brain 5 . Moreover, a recent study showed that visual stimulation can enhance CSF-flow in the 4 th ventricle 25 . To investigate whether visual stimulation also drives local CSF-mobility in the SAS and PVS of the visual area in humans, a flashing checkerboard stimulus was shown to nine individuals while measuring CSF-mobility. Entraining vasomotion by means of 0.1 Hz visual stimulation significantly increased CSF-mobility in the visual cortex by 1.2 % on average (range = 0.2 - 3.4 %, Wilcoxon signed rank test p=0.004, Fig. 7A-B), compared to a control region outside of the stimulated area (average difference in control region = 0.01 %, range = -0.35 - 0.36 %, n=9, Wilcoxon signed rank test p=0.82). Note that the “visual cortex” region consists of the CSF within the activated area in response to the visual stimulation (see methods). To investigate whether the spread in increase in CSF-mobility across individuals originated from different responses to visual stimulation, we calculated the effect of visual stimulation on evoked vascular reactivity by quantifying the BOLD signal amplitude change. A Bayesian correlation analysis gave anecdotal evidence for a positive correlation between BOLD amplitude and CSF-mobility change (BF 0+ =2.3, Fig. 7C), suggesting that the vessel wall displacement induced by visual stimulation potentially leads to the increase in CSF-mobility 28 . The CSF-mobility increase induced by entraining vasomotion approaches the amplitude change driven by cardiac and respiratory cycles (Fig. 7D). Regional alterations of CSF-mobility and FA in patients with CAA Lastly, to show the potential of CSF-STREAM to improve our understanding of neurodegenerative diseases, we applied the MRI-sequence in eight patients with a clinical diagnosis of sporadic CAA, and in eight age- and sex-matched healthy controls. We found a 20% increase in CSF-mobility in the SAS closely surrounding the MCA of CAA patients as compared to healthy controls (CSF-mobility CAA = 0.042±0.006 mm 2 /s, CSF-mobility Control = 0.035±0.005 mm 2 /s, p=0.01, Mann-Whitney U-test, Fig. 8B). This increase in CSF-mobility was accompanied with a 10% decrease in FA (FA CAA = 0.65±0.06 mm 2 /s, FA Control = 0.73±0.04 mm 2 /s, p=0.02, Mann-Whitney U-test, Fig. 8C). Of note, this significant group difference was detectable in the SAS around the MCA up to a radius of approximately 1.7 mm around the vessel; beyond this distance, the difference diminished (Extended data Fig. 7). In PVS around penetrating vessels in the centrum semiovale (CSO), CSF-mobility and FA were not significantly different between groups (CSF-mobility CAA = 0.016±0.006 mm²/s, CSF-mobility Control = 0.017±0.006 mm²/s, p ≥ 0.05, Mann-Whitney U-test), whereas a significantly increased PVS volume was found in CAA patients compared to controls, as expected 29 (Fig. 8H, p = 0.007, Mann-Whitney U-test). Age, sex, lifestyle factors as well as cognitive and physical health scores did not differ between both groups (Extended data Fig. 8). Discussion In this study we measured and directly compared the effects induced by various driving forces not only in large CSF-spaces such as the 4 th ventricle or in the SAS at the brain surface, as was shown by previous studies 4,24–26,30 , but also for the first time in smaller PVS around penetrating arteries, which are closer to waste production sites. Moreover, we provide the first in vivo evidence of altered CSF-mobility in patients with CAA, a disease associated with brain clearance impairment. A strength of our results in the younger, healthy cohort is that a direct comparison can be made between the cardiac/respiration/random cycles as they all originate from the same dataset, i.e. one dataset was reconstructed in different ways, thus capturing the individual in the same physiological state. Our study shows that driving forces for CSF-mobility are region specific. In large CSF-spaces located at the base of the brain (SAS around the MCA and 4 th ventricle), the cardiac cycle appears to be a larger contributing driving force to CSF-mobility as compared to respiration. This is in line with previous findings in mice 4 and humans 23,30 and can be explained by the fact that during systole the brain expands due to increased filling of its entire arterial system, which leads to CSF flowing out of the skull to compensate for the increased brain volume. This flow reverses during diastole. Conversely, in PVS, located around penetrating arteries more distally in the vascular tree where cardiac pulsations are likely dampened down, our results point to similar influence on CSF-mobility fluctuations by cardiac and respiratory cycles. This new finding in PVS suggests both cardiac and respiratory rhythms are equally important driving forces for perivascular CSF-mobility. This would support the notion of multiple sources promoting mixing within PVS 12 , as opposed to a single wave caused by a single physiological phenomenon (e.g. only arterial pulsations) pushing CSF through PVS 4,11 . It should, however, be understood that in our study the six cardiac and respiratory phases originate from data collected over 40 minutes. Therefore, possible effects from different types of respiration (e.g. single events of deep breaths) would be averaged out. It can be hypothesized that forced breathing or deep inspirations could further enhance CSF-mobility fluctuations, as shown previously 24,30 . Low-frequency (~0.1 Hz) vasomotion has also been proposed as a driving force for brain clearance, and was previously experimentally shown to be associated with increased tracer movement in CSF 5,11 . The influence of resting-state vasomotion can, unfortunately, not be assessed by retrospective binning as it was done with the cardiac and respiratory cycles, since it does not provide an accessible external triggering opportunity. Animal experiments have shown that low frequency visual stimulation can be used to entrain vasodilation at the vasomotion frequency and thereby speed up clearance of fluorescent-labeled tracers 5 . Moreover, in humans it was recently shown that a visual stimulation could drive global CSF-flow in the 4 th ventricle 25 . By using a similar stimulation strategy to entrain vasomotion, we showed that CSF-mobility within the visual cortex area itself is significantly enhanced in humans during 0.1 Hz frequency visual stimulation and that the amplitude of this effect approaches the strength of cardiac and respiratory fluctuations. These findings might even lead to the provocative hypothesis that, similar as in rodents, a simple visual stimulation paradigm at an optimal frequency may be able to enhance brain clearance in the visual cortex 31 . The ability of CSF-STREAM to measure CSF-mobility within the stimulated region, i.e. much closer to the production sites of metabolic ‘waste’ products, should be considered an important advantage over techniques that measure CSF-flow at a single location at the exit of the brain, i.e. the aqueduct or the 4 th ventricle. The observed increase in CSF-mobility by entraining vasomotion might mirror similar, albeit stronger, physiological fluctuations which occur during sleep. Indeed, in sleep, low frequency oscillations in cerebral blood volume, measured using BOLD fMRI 26,32 , are of similar amplitude as the ones measured in our study during a 0.1 Hz visual stimulation (Fig. 7C). As these oscillations are brain-wide during sleep, and not localized to a stimulated region, this could in turn lead to brain-wide enhancement of CSF-mobility during sleep. While the increase in CSF-mediated clearance during sleep is proposed to be a key feature of its proper function, clearance failure has been suggested to be part of the pathogenesis of CAA, a common small vessel disease and contributor to dementia, characterized by vascular amyloid-β accumulation and subsequent brain lesions such as hemorrhages. Our pilot study in CAA patients provides the first in vivo confirmation of observations made in rodent models of CAA. We observed regional alterations in CSF-mobility and FA in patients with sporadic CAA compared to healthy controls, even in our relatively small sample size: CSF-mobility was found to be increased by 20% and FA decreased by 10% within the SAS surrounding the MCA, whereas these parameters remained unchanged in PVS. Our findings are in line with the observations in the rodent model of CAA, where a similar 20% increase in CSF-speed was found in the SAS at the skull base, whereas CSF-speed in tissue remained unchanged 19 . The higher CSF-mobility together with lower FA might indicate higher but more disorganized CSF-mobility patterns in the SAS with CSF bypassing the tissue compartment. Such bypassing might be caused directly by amyloid-β accumulation in cortical or leptomeningeal vessels or indirectly by stiffening of arterioles and loss of smooth muscle cells 33 . Interestingly, we found that the distance of the CSF-volume to the vessel is of importance in the SAS surrounding the MCA (Extended data Fig. 7): when measuring in CSF located within ~1.7 mm from the MCA, CSF-mobility and FA were found different between CAA and healthy controls, whereas when including CSF located further away from the MCA, the changes between groups attenuated. This might be explained by an additional membrane in the SAS along the major cerebral arteries that compartmentalizes the SAS, as recently proposed by Eide and Ringstad 34 . The newly introduced CSF-STREAM technique was inspired by the long echo-time diffusion tensor imaging approach (DTI) proposed by Harrison and colleagues to measure CSF-mobility in the SAS surrounding the MCA in the rat brain 35 . The isolation of CSF-signal is essential to the successful measurement of CSF-mobility in PVS, as it ensures that the measured signal solely originates from CSF and not from slow-flowing blood in the vessel located inside of the PVS. A good separation of the CSF-signal was achieved, as shown in Fig. 1, Extended data Fig. 1 and Suppl. video 1. The use of motion-sensitizing gradients applied during T 2 -preparation instead of a traditional diffusion sequence enabled the robust use of a multi-shot turbo-spin-echo (TSE) readout for which the image quality and the achievable resolution is significantly higher than with an EPI readout that is usually the backbone of DTI-scans: the motion information encoded in the T 2 -preparation is stored in the longitudinal plane and therefore avoids phase accruals that would make the multi-shot reconstruction subject to image artefacts. Altogether, using the proposed CSF-STREAM technique, we showed that CSF-mobility could be measured non-invasively and at a high spatial resolution in healthy individuals and patients by combining accelerated ultra-high-field MRI while exploiting the magnetic properties of CSF and motion-sensitizing gradients. The measured CSF-mobility values are more than ten times higher than water diffusion values, even in PVS (~1×10 -3 mm 2 /s for water diffusion in the brain tissue 36 versus 10-50×10 -3 mm 2 /s for CSF-mobility). This high CSF-mobility combined with a high FA indicates that the physiological process behind CSF-mobility is not pure diffusion or plug flow, but more likely laminar flow or back-and-forth motion 22 . This is further substantiated by the influence of physiology on CSF-mobility: if the measured CSF-mobility was diffusion-dominant, then no effect of the cardiac or respiratory cycles should have been observed 22,23,27 . The motion sensitizing used in this study was 3.5 mm/s, which was sufficient to partially attenuate the CSF-signal in PVS, indicating that a part of CSF in PVS moves faster than 3.5 mm/s (Suppl. Fig. 1). Since the current sequence does not allow for measuring phase-accrual generated by the motion-sensitizing gradients, it is not possible to determine whether CSF exhibits net flow in the PVS or if the CSF-mobility in PVS would be better described as a mixing phenomenon: the direction (towards one end of the axis of movement or the other, e.g. right-to-left or left-to-right) cannot be determined with the current sequence. Still, a potential propagation pattern of CSF-mobility across cardiac phases can be observed (Fig. 3, Fig. 4, Suppl. video 2, Suppl. video 3), suggesting that CSF-mobility varies in distinct spatially coherent waves driven by the cardiac and respiratory cycle. This could further enhance the efflux of waste products from the neuropil to the SAS. A further limitation of the current approach is the long scan duration that leads to a higher sensitivity to motion (two datasets had to be excluded due to motion artefacts) and to a low intrinsic temporal resolution, which limits the study of pulsatile dynamics that cannot be predicted in advance or using an external trigger (e.g. vasomotion, as opposed to cardiac or respiratory dynamics). Regarding the scan duration, the current k-space sampling rendered sufficient SNR, allowing to further split k-space to study the effect of cardio-respiratory pulsations. This suggests that for baseline CSF-mobility measurements (i.e. without retrospective binning), further acceleration can be achieved. Shorter acquisition times resulting from such higher acceleration combined with the use of prospective motion correction, e.g. by navigators 37,38 , would also help reduce sensitivity to motion. However, even with the current scan-times it was possible to identify CSF-mobility changes in patients with a neurodegenerative disease. The truly non-invasive nature of the developed technique (as opposed to techniques using contrast agent injections) allows the inclusion of this methodology into longitudinal, patient, and population studies. It also makes the sequence repeatable (i.e. applicable several times in a row), which would be essential when applied to sleep research or to monitor disease progression for example in neurodegenerative diseases. The information obtained from such studies could lead to important insights into the ways human brain pathologies are affected by impaired brain clearance and lead to new ways of improving brain health. Conclusion CSF-STREAM enables detailed measurement of CSF-mobility in humans, from large CSF-filled spaces down to PVS surrounding penetrating arteries in a fully non-invasive manner. Cardiac and respiratory fluctuations induced comparable oscillations in PVS, whereas in larger CSF-spaces at the base of the brain, the cardiac cycle was found to be the main driving force of CSF-mobility. Moreover, regional alterations in CSF-mobility were found in patients with a presumed brain clearance disorder. Finally, CSF-mobility in the visual cortex could be enhanced through entraining vasomotion at 0.1 Hz. Methods Subjects Healthy, younger cohort A total of 24 healthy individuals (age: 33±13 years, 20 females, 4 males) were scanned: 14 individuals were enrolled in the first study to evaluate CSF-mobility fluctuations across cardiac/respiration/random phases. One individual was excluded because of motion artefacts and two because of insufficient quality of the cardiac signal. Motion was identified in the reconstructed images as blurring of the images and duplication of brain structures. Ten participated in the second study to investigate the effect of a visual stimulation on CSF-mobility, of which one was excluded due to motion artefacts. Six individuals participated in both studies. All were screened for MRI contra-indications and provided written informed consent. All experiments were performed in accordance with the Leiden University Medical Center Institutional Review Board (Leiden, The Netherlands) under authorization number P07.096. CAA cohort CAA patients (n=8) were recruited through the neurovascular outpatient clinic at University Hospital Bonn. The diagnosis of probable CAA was independently confirmed by a board-certified neuroradiologist according to the Boston Criteria version 2.0 29 . Age- and sex-matched healthy participants (n=8) were recruited through the DANCER cohort, i.e. a neurologically unaffected control cohort of the German Center for Neurodegenerative Diseases (DZNE). The study was approved by the Ethics Committee of University Hospital Bonn. Written informed consent was obtained for each participant. Medical history was obtained from each participant to record relevant previous diseases, cardiovascular risk factors, degree of disability, lifestyle factors and physical activity. Degree of disability was assessed using the modified Rankin Scale. Physical activity was assessed using the physical activity scale for the elderly (PASE). Global cognitive status was assessed using the Montreal cognitive assessment test. Trail Making Tests (TMT) A and B were used to assess executive function and processing speed. Symptoms of depression were assessed using the Center for Epidemiologic Depression Scale (CES-D), revised Becker Depression Inventory (BDI-II) and Geriatric Depression Scale (GDS). MRI scans acquisition Healthy, younger cohort All scans were acquired using a 7 Tesla MRI scanner (Achieva, Philips, Best, The Netherlands) equipped with a quadrature birdcage head coil and a 32-channel receive coil array (Nova Medical, Wilmington, MA, USA). Anatomical 3D-T 1 scan : 3D T 1 -weighted images were acquired using the following parameters: field-of-view = 246×246×225 mm 3 , flip angle = 7 ◦ , echo time (TE) = 1.9 ms, repetition time (TR) = 4.2 s, spatial resolution = 0.9 mm isotropic, and acquisition time = 142 s. CSF-STREAM : High-resolution, whole-brain (0.45 mm isotropic voxel-size, field-of-view = 250×250×190 mm), 3D images were acquired with a TSE sequence (TE = 495 ms, TR = 3.4 s, TSE-factor 146, excitation & refocusing FA = 90 ◦ ). This long echo-time readout was combined with a T 2 -preparation module (duration = 37 ms, 2 refocusing pulses) in order to allow the insertion of motion sensitizing gradients and to further isolate the CSF-signal. To accelerate the acquisition, k-space undersampling in ky and kz phase encoding directions compatible with compressed sensing reconstruction was performed using the Amsterdam UMC PROUD patch 39 , based on a pseudo-radial variable density (density decay = 0.5) sampling pattern with a fully sampled 29×29 auto-calibration area in the center of k-space. This way, an acceleration factor of 17 was achieved, which allowed to obtain high spatial resolution, whole-brain, static CSF-images in 5 min 30 s. Subsequently, motion-sensitizing gradients were included in the T 2 -preparation in order to encode CSF-mobility. Seven sets of volumes (sub-scans) were acquired: one without motion-sensitizing gradients and six with gradients applied in different, orthogonal directions. In practice, motion-sensitized gradients of 5 mm/s were played out on two axes simultaneously, resulting in a diagonal direction with a motion-encoding of 5/√2 = 3.5 mm/s. The acquisition time per sub-scan was 5 min 30 s, yielding a total scan time of 38 min 30 s. Visual stimulation scout fMRI scan : A visual scout fMRI blood oxygen level dependent (BOLD) scan was acquired in order to locate the visual cortex. The acquisition parameters were as follows: field-of-view = 222×190 mm 2 , flip angle = 70 ◦ , TE = 22 ms, TR = 2 s, 35 slices, voxel size: 1.97×1.74 mm 2 in-plane, 2 mm slice thickness, EPI-factor = 43, 60 repetitions (timepoints), acquisition time = 128 s. The visual stimulus consisted of 3 blocks of an 8 Hz flashing radial black-and-white checkerboard pattern for 20 seconds alternated with 20 seconds of a fixed grey screen as rest condition. Using the above-mentioned scans, two studies were performed in the healthy, younger cohort: Study 1: CSF-mobility across cardiac, respiration and random phases – This protocol consisted of an anatomical 3D-T 1 scan, a CSF-STREAM scan (one scan = one non-motion-sensitized sub-scan + 6 motion-sensitized sub-scans) and a visual stimulation scout scan (except for the first four subjects). During the acquisition, the heart rate was continuously recorded using a peripheral pulse unit and the respiration rate using a belt wrapped around the subject’s chest; both physiological monitoring devices were the standard equipment as supplied by the vendor (Philips, Best, The Netherlands). Study 2: entrained vasomotion using a 0.1 Hz visual stimulation – This protocol consisted of an anatomical 3D-T 1 scan, a CSF-STREAM scan and a visual stimulation scout scan. During each of the seven CSF-STREAM sub-scans, a visual stimulus was shown during the first half of the sub-scan, and a grey screen during the second half of the sub-scan. The stimulation paradigm consisted of an 8 Hz flashing radial black and white checkerboard pattern for 5 s alternated with 5 s of a fixed grey screen, altogether leading to a 0.1 Hz stimulation frequency. In order to sample k-space homogeneously in both sub-scan halves, k-space sampling was readjusted by first acquiring the odd TSE-shots of the original k-space sampling in the first half and subsequently the even TSE-shots during the second half. CAA cohort MRI data were acquired at a 7 Tesla MRI system (Siemens Healthineers, Erlangen, Germany) using a head array coil with 32 receive and 8 transmit channels (Nova Medical, Wilmington, MA, USA). Scans were performed in the morning between 09:30 am and 11:30 am. The scan protocol included a T 1 -weighted multi-echo MPRAGE (0.80 mm isotropic, TR = 2800 ms, TI = 1100 ms, scan time = 4 min) and a CSF-STREAM scan (0.50 mm isotropic, TE = 515 ms, TR = 3400 ms, TSE-factor 146, refocusing FA = 70 ◦ , 12x Poisson-disk undersampling scheme, acquisition time 4:15 min per sub-scan, motion-sensitizing gradients of 4 mm/s). In practice, similar to the study in healthy, young volunteers, motion-sensitized gradients of 5.6 mm/s were played out on two axes simultaneously, resulting in a diagonal direction with a motion-encoding of 4 mm/s. Additionally, susceptibility-weighted images (0.50 mm isotropic) were acquired to quantify cerebral microbleeds and to identify superficial siderosis. CSF-STREAM image reconstruction Healthy, younger cohort All CSF-STREAM reconstructions of this cohort were performed offline in MATLAB 2018b (The Mathworks, USA), using an in-house built reconstruction pipeline developed within ReconFrame (GyroTools, Zürich, Switzerland) in combination with the open-source Berkeley Advanced Reconstruction Toolbox (BART) 40 version 0.4.03. Study 1: retrospective cardiac, respiration and random binning - After acquisition, each of the seven CSF-STREAM sub-scans was reconstructed three times: the k-space profiles were retrospectively binned in 6 phases, using retrospective binning to either (1) the cardiac cycle, (2) the respiratory cycle or (3) to random phases (negative control). The R-R cardiac peaks as well as the respiration peaks were detected automatically from the recorded cardiac and respiratory traces using a MATLAB script (using the findpeaks function) and the results were checked manually for each scan and corrected if wrongly detected. Random phases were generated using the randi MATLAB function. The retrospective binning in 6 phases was performed in two steps in order to preserve image quality: k-space was binned twice in 3 phases, with a 1/6 th phase shift between the two steps, and subsequently combined in one dataset (Extended data Fig. 3). The T 2 -preparation preceding each TSE-shot was taken as reference and the signal in the subsequent TSE-shot was considered to be dependent on this specific cardiac/respiration/random phase 41,42 . Each reconstruction step was performed using BART’s pics command with total variation in the temporal domain with a regularization factor of 0.005 and input coil sensitivities estimated from the k-space center using BART’s caldir command. Altogether, this generated 6 tensors per driving force per subject (i.e. one tensor per cardiac/respiration/random phase). Study 2: 0.1 Hz visual stimulation – Thetwo half k-spaces (0.1 Hz stimulation versus rest) of each of the seven CSF-STREAM sub-scans were reconstructed using BART’s pics command with total variation in the temporal domain with a regularization factor of 0.005 and input coil sensitivities estimated from the k-space center using BART’s caldir command. Per subject, this resulted in a set of two tensors: one with 0.1 Hz stimulation and one at rest. CAA cohort Image reconstruction of CSF-STREAM sub-scans was achieved using the pics command from the BART Toolbox with l1-regularization (regularization factor 0.002) and 30 iterations. Input coil sensitivities were estimated from a fully-sampled gradient-echo pre-scan using BART’s ecalib command. Post-processing – healthy, younger cohort CSF-mobility, FA and principal orientation of CSF-mobility Intra-subject images were registered using Elastix version 4.9.0 43 . CSF-mobility, its principal orientation, and FA were modelled using a MATLAB script 44 by computing the mean eigenvalue of a rank-two positive definite tensor, analogous to DTI. CSF-mobility is a measure of the amount of movement that CSF undergoes within a given time in a voxel as a function of the intra-voxel dephasing of signal due to the application of bipolar gradients. It is measured in units of mm 2 /s and is calculated in a similar way as an apparent diffusion coefficient. For the first study that investigated the effect of the cardiac/respiration/random pulsations, six CSF-mobility and FA maps were created (one for each phase) per driving force per subject. CSF-mobility change and FA-change maps were obtained for each cardiac, respiration and random phase after normalizing the maps voxel-wise to the mean value over phases. CSF-mobility and FA changes across driving forces were investigated in six ROIs. As can be seen in Fig. 3 and Fig. 4, the fluctuations across phases vary spatially, i.e. the phase of maximum signal is location-dependent. To avoid phase cancellation within an ROI and hence potential smoothing of the effect of a driving force, the time-profiles were realigned voxel-wise before calculating the average CSF-mobility and FA changes in an ROI. That way, phase 1 always contained the maximum signal change and was discarded from the plots in Fig. 5 and Extended data Fig. 4. To quantitatively compare how CSF-mobility varies across the proposed driving forces, the original CSF-mobility change over 6 phases (i.e. without the realigment mentioned in the previous paragraph) was fitted voxel-wise to a sine function. The fit quality, the maximum amplitude as well as the phase of the maximum amplitude were the fitted parameters. The amplitude of voxels with a low fit quality (R 2 <0.5) was set to 0 to better represent the absence of coherent change across phases. For example, in the right panel of Extended Fig. 6, the fitted curve should ideally be flat (i.e. amplitude of 0% instead of 12%) because the signal pattern is more noisy (“zig-zag”) than coherent across phases. For the second study investigating the effect of entrained vasomotion on CSF-mobility, two CSF-mobility maps were created: one acquired during the 0.1 Hz visual stimulation and another at rest. The relative difference between the average CSF-mobility during the 0.1 Hz stimulation and rest (no stimulation) was computed to investigate the effect of entrained vasomotion. In the subjects who participated in both studies, the effect induced by entrained vasomotion was compared to that from cardiac, respiratory, and random cycles (Fig. 7D) by computing the difference between the maximum and minimum CSF-mobility change over cardiac/respiration/random phases. All results were visualized using MATLAB and Paraview 45 . When plotting CSF-mobility/FA volume renderings or CSF-mobility change maps, voxels where the CSF-signal was lower than 150 were masked to exclude noise. The CSF-mobility orientation plots were visualized in smaller volumes of interest. Before computing the CSF-mobility orientation as described above, the resolution of the small volume of interest of the seven CSF-STREAM sub-scans was doubled using the MATLAB function imresize3 , in order to better show the orientation in the selected regions. fMRI processing z-score map generation : fMRI data processing was carried out using FEAT (FMRI Expert Analysis Tool) Version 6.00, part of FSL. The following pre-statistics processing were applied: motion correction using MCFLIRT 46 ; slice-timing correction using Fourier-space time-series phase-shifting; non-brain removal using the Brain Extraction Tool (BET) 47 ; spatial smoothing using a Gaussian kernel of FWHM 3 mm; grand-mean intensity normalization of the entire 4D dataset by a single multiplicative factor; high-pass temporal filtering (Gaussian-weighted least-squares straight line fitting, with sigma = 50s). To investigate the possible presence of unexpected artefacts or activation, ICA-based exploratory data analysis was carried out using MELODIC 48 . The statistical analysis of the time-series was carried out using FILM with local autocorrelation correction 49 . z-score template (study 1): As the first four subjects of study 1 did not have a visual stimulation scout scan, a mean z-score template was generated to create a visual cortex mask for study 1, using the scout scan of a subset of ten subjects (from study 1 and 2; only one scout scan per subject was included, i.e. scans from the six subjects included in both studies were only included once). To create this z-score template, the ten BOLD scans were registered to the 3D-T 1 scans using a boundary based registration 50 . This transformation was applied to the z-score maps. Then, 3D-T 1 scans were registered to MNI-space using FSL’s FLIRT 46,51 and FNIRT 52 . The resulting warpfield was applied to the registered z-score maps. The template was then generated using a one-sample group mean generalized linear model within FreeSurfer 53 and transformed into the CSF-mobility space of each individual subject of study 1. Individual z-score map (study 2): For study 2, a visual stimulation scout scan was available for each subject. Therefore, for this study, individual z-score maps were used to best detect the visual cortex in each subject. The person-specific z-score maps were thus registered to each individual CSF-mobility space in the following way. First, the BOLD scans were registered to the 3D-T 1 scans using a boundary based registration 50 . Subsequently, the 3D-T 1 scans were skull stripped and segmented in brain tissue types 54 . The obtained CSF probability map was registered to CSF-mobility space with an Euler registration using Elastix 43 . The resulting transformations were then applied to the z-score maps. The obtained z-score maps were used to detect the visual cortex. The BOLD signal amplitude was calculated from the registered, motion corrected, slice-time corrected BOLD images. The 3 stimulation patterns (3×20 dynamics) were first averaged together, then the relative difference between baseline signal (averaged over dynamics 5-10) and maximum signal (averaged over dynamics 15-20) was computed. ROI definition ROIs were manually delineated using anatomical landmarks on the non-motion-sensitized CSF-scan (cf. Fig. 2) using ITK-Snap 55 , as follows: The 4 th ventricle ROI was drawn over 13 transversal slices The ROI delimiting the SAS around the MCA was drawn over seven transversal slices on the left MCA branch The motor cortex SAS sulci ROI was drawn over ten transversal slices PVS in the basal ganglia were identified over sagittal slices as CSF-filled spaces around the lenticulostriate arteries The ROI of PVS surrounding penetrating arteries in the white matter was drawn over 25 transversal slices in the centrum semiovale, starting from the slice directly above the lateral ventricles The blood ROI was drawn inside one MCA branch over two slices The noise ROI was drawn outside the brain, on the two central sagittal slices and on the corners of the same slice where the SAS MCA ROI was drawn For study 1, the visual cortex ROI was defined based on the template z-score output (four datasets of study 1 did not contain a visual stimulation scout scan). A threshold of z-score>3.5 was used for all subjects of this study. Only the largest cluster of contiguous voxels was included in the mask. As the resolution of the fMRI scan used to create the z-score map was much lower than that of the CSF-STREAM, the obtained area not only contained the visual cortex but also the CSF in its vicinity. To extract the final areas of interest, the manually delineated ROIs and the visual cortex ROI were multiplied with a CSF-mask, thus including only voxels containing CSF and not noise. This CSF-mask was created by first thresholding the non-motion-sensitized CSF-scan using a threshold of 150 (a.u.). After visual inspection, this threshold could be adapted individually to assure proper selection of PVS. For study 1, if a voxel had a CSF-mobility change higher than 50% in one or more of the cardiac/respiration/random datasets, it was excluded from the mask. Voxels that had no included neighboring voxels (“lonely” voxels) were also excluded from the mask. For the two PVS ROIs, an additional Frangi filter 56 (0.6<σ<1 with a step of 0.2, Frangi vesselness constant = 0.5) was applied in order to ensure the exclusive inclusion of vessel-like structures and exclusion of noise. For study 2, a visual stimulation scout scan was available for all subjects, and we therefore used individual z-score maps to create the visual cortex ROIs. A threshold of z-score>7 was used to define the visual area and a threshold of z-score<1 was used to define a control region (rest of the brain). For the visual cortex ROI, only the largest cluster of contiguous voxels was included in the ROI. Next, to ensure only voxels containing CSF were included and not noise, a mask based on the non-motion-sensitized CSF-scan was created using a threshold of 150 (a.u.). Voxels for which the CSF-mobility change between the 2 conditions (stimulation ON and OFF) was higher than 50% were considered as noise and excluded from the ROI. As the resolution of the fMRI scan used to create the z-score map was much lower than that of the CSF-STREAM, the obtained area not only contained the visual cortex but also the CSF around the visual cortex (SAS and PVS). The effect induced by the cardiac and respiratory cycles was compared to the changes induced by entrained vasomotion in the subjects who participated in both studies. To that end, the personalized visual area mask created for study 2 was registered to the CSF-space of study 1 using Elastix. Post-processing - CAA cohort CSF-mobility and FA The CSF-STREAM sub-scans were first interpolated from a 0.50 mm to a 0.17 mm isotropic resolution and co-registered using Elastix. Subsequently, the mean eigenvalue of a rank-two positive definite tensor was computed (DTI post-processing) using Python 3.10to assess CSF-mobility and FA. To minimize the effects of background noise, a cut-off value of 0.15 mm²/s was used to exclude all voxels with unphysiologically high CSF-mobility values. ROI definition SAS-MCA segmentations: the M1 segment of the middle cerebral artery was first segmented semi-manually using anatomical landmarks on the non-motion-sensitized CSF-scan using MITK version v2022.10. The MCA-segmentation was then inflated using the imdilate Matlab function to create a CSF-mask containing the SAS around the MCA; inflation was done from 0.17 mm up to 3.00 mm with a step-size of 0.17 mm. To ensure only CSF-signal was selected in the mask, voxels with low CSF-signal in the non-motion sensitized scan were excluded. PVS segmentations: Parcellated atlases from the T 1 -scan were generated using Freesurfer (Version 6.0). White matter segmentations, derived from the Freesurfer parcellation, were manually corrected if necessary. To capture comparable ROIs of the CSO in all participants, the eyes and the optic chiasm served as anatomical landmarks for reference plane definition. The dimensions of the CSO segmentation encompassed the entire white matter above the lateral ventricles. PVS within the defined CSO segmentation were semiautomatically segmented using a Meijering filter-based approach 57 with a global threshold on the interpolated non-motion sensitized scan of CSF-STREAM (0.17 mm isotropic). Assessment of microbleeds Cerebral microbleeds were quantified and superficial siderosis was identified by a board-certified neuroradiologist with 8 years of experience according to the STRIVE-2 rating scale. Statistical analysis Within each ROI, we evaluated whether the fit quality as well as the amplitude of CSF-mobility change were significantly different between driving forces. A post-hoc Bonferroni-corrected pairwise comparison between driving force effects was performed using a Wilcoxon signed rank test when Friedman’s test was significant. To evaluate the effect of the visual stimulation, a Wilcoxon signed rank test was performed on the average CSF-mobility values with and without stimulation from each subject. These above-mentioned statistical analyses were performed in MATLAB. To evaluate the evidence for a correlation between the BOLD amplitude and CSF-mobility change with a visual stimulation, a Bayesian correlation analysis was performed using JASP software 58 (JASP Team (2022). JASP (Version 0.16.4)). A stretched beta prior with a width of 1.0 was used, and the Bayes Factor was tested to be stable over a range of prior settings. To evaluate differences in CSF-mobility, FA, ROI volume and patient information between CAA and healthy controls, Mann-Whitney U-tests were performed. Unless mentioned otherwise, shaded error areas and error bars represent confidence intervals of SD×1.96/√n, n being the number of included subjects. Boxplots were plotted using the MATLAB boxplot function: on each boxplot, the central line indicates the median, and the bottom and top edges of the box indicate the 25 th and 75 th percentiles, respectively. The whiskers extend to the most extreme data points not considered outliers, and the outliers are plotted individually using the '+' marker symbol. Preliminary parts of this work were presented at conferences 41,42,59,60 . Declarations Acknowledgements: The authors would like to thank Lukas Gottwald and Aart Nederveen for providing the Amsterdam UMC PROUD patch, Thomas Roos and Kai Lønning for help with the reconstructions, Geir Ringstad for fruitful discussions and Rüdiger Stirnberg for his support and input with regards to the 7 Tesla MRI protocol. This work is part of the research program Innovational Research Incentives Scheme Vici with project number 016.160.351, which is financed by the Netherlands Organization for Scientific Research (NWO). It was furthermore supported by the Women in MR award from the ISMRM-Benelux, Alzheimer Nederland (Young Outstanding Researcher Award WE.25-2020-05 to Susanne van Veluw and travel grant to Lydiane Hirschler), the Joint Program for Neurodegenerative Diseases (JPND) on Human Brain Clearance Imaging (HBCI), the Federal Ministry of Education and Research (BMBF) in Germany (funding code 01KX2130), as well as the Leducq Foundation (Transatlantic Network of Excellence 23CVD03). Katerina Deike was funded by the Medical Faculty of University Bonn (2022-FKS-02, 2024-FKS-02 and Femhabil 03-2022). Gabor Petzold received funding from the DZNE and Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy – EXC2151 – 390873048. Daniel Paech was funded by a grant (2022-EKES.33) of the Else Kröner-Fresenius-Stiftung (EKFS), Philipp Vollmuth is supported through an Else Kröner Clinician Scientist Endowed Professorship by the Else Kröner Fresenius Foundation (reference number: 2022_EKCS.17) and Alexander Effland was funded by the German Research Foundation under Germany’s Excellence Strategy EXC-2047/1–390685813 and EXC2151–390873048. Conflicts of Interest: The LUMC receives research support from Philips, M.W.A. Caan is a shareholder of Nicolab International Ltd, Katerina Deike and Daniel Paech are co-founders and shareholders of relios.vision GmbH. Daniel Paech is on the Guerbet advisory board. Data availability: Data will be made available upon reasonable request and with a data exchange agreement. Code availability: Codes are available upon reasonable request. References Dolgin, E. Brain’s drain. Nat. Biotechnol. (2020) doi:10.1038/s41587-020-0443-1. Rasmussen, M. K., Mestre, H. & Nedergaard, M. Fluid Transport in the Brain. Physiol. Rev. (2021) doi:10.1152/physrev.00031.2020. Iliff, J. J. et al. A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid β. Sci. Transl. Med. (2012) doi:10.1126/scitranslmed.3003748. Mestre, H. et al. Flow of cerebrospinal fluid is driven by arterial pulsations and is reduced in hypertension. Nat. Commun. (2018) doi:10.1038/s41467-018-07318-3. van Veluw, S. J. et al. Vasomotion as a Driving Force for Paravascular Clearance in the Awake Mouse Brain. Neuron (2020) doi:10.1016/j.neuron.2019.10.033. Ringstad, G., Vatnehol, S. A. S. & Eide, P. K. Glymphatic MRI in idiopathic normal pressure hydrocephalus. Brain (2017) doi:10.1093/brain/awx191. Ringstad, G. et al. Brain-wide glymphatic enhancement and clearance in humans assessed with MRI. JCI insight (2018) doi:10.1172/jci.insight.121537. Wardlaw, J. M. et al. Perivascular spaces in the brain: anatomy, physiology and pathology. Nat. Rev. Neurol. (2020) doi:10.1038/s41582-020-0312-z. Bakker, E. N. T. P., Naessens, D. M. P. & VanBavel, E. Paravascular spaces: entry to or exit from the brain? Exp. Physiol. (2019) doi:10.1113/EP087424. Mestre, H., Mori, Y. & Nedergaard, M. The Brain’s Glymphatic System: Current Controversies. Trends Neurosci. xx , 1–9 (2020). Aldea, R., Weller, R. O., Wilcock, D. M., Carare, R. O. & Richardson, G. Cerebrovascular smooth muscle cells as the drivers of intramural periarterial drainage of the brain. Front. Aging Neurosci. 11 , (2019). Zhao, L., Tannenbaum, A., Bakker, E. N. T. P. & Benveniste, H. Physiology of Glymphatic Solute Transport and Waste Clearance from the Brain. Physiology (Bethesda). 37 , 0 (2022). Vinje, V. et al. Respiratory influence on cerebrospinal fluid flow – a computational study based on long-term intracranial pressure measurements. Sci. Rep. (2019) doi:10.1038/s41598-019-46055-5. Rasmussen, M. K., Mestre, H. & Nedergaard, M. The glymphatic pathway in neurological disorders. The Lancet Neurology (2018) doi:10.1016/S1474-4422(18)30318-1. Greenberg, S. M. et al. Cerebral amyloid angiopathy and Alzheimer disease — one peptide, two pathways. Nature Reviews Neurology (2020) doi:10.1038/s41582-019-0281-2. Christensen, J., Wright, D. K., Yamakawa, G. R., Shultz, S. R. & Mychasiuk, R. Repetitive Mild Traumatic Brain Injury Alters Glymphatic Clearance Rates in Limbic Structures of Adolescent Female Rats. Sci. Rep. 10 , (2020). Mestre, H. et al. Cerebrospinal fluid influx drives acute ischemic tissue swelling. Science (80-. ). (2020) doi:10.1126/science.aax7171. van Veluw, S. J. et al. Is CAA a perivascular brain clearance disease? A discussion of the evidence to date and outlook for future studies. Cell. Mol. Life Sci. 81 , (2024). Chen, X. et al. Cerebral amyloid angiopathy is associated with glymphatic transport reduction and time-delayed solute drainage along the neck arteries. Nat. Aging 2 , (2022). Albargothy, N. J. et al. Convective influx/glymphatic system: tracers injected into the CSF enter and leave the brain along separate periarterial basement membrane pathways. Acta Neuropathol. (2018) doi:10.1007/s00401-018-1862-7. Janssen, P. M. L., Biesiadecki, B. J., Ziolo, M. T. & Davis, J. P. The need for speed: Mice, men, and myocardial kinetic reserve. Circ. Res. 119 , (2016). Bito, Y., Harada, K., Ochi, H. & Kudo, K. Low b-value diffusion tensor imaging for measuring pseudorandom flow of cerebrospinal fluid. Magn. Reson. Med. 86 , (2021). Wen, Q. et al. Assessing pulsatile waveforms of paravascular cerebrospinal fluid dynamics using dynamic diffusion‐weighted imaging (dDWI). Neuroimage 260 , 119464 (2022). Töger, J. et al. Real-time imaging of respiratory effects on cerebrospinal fluid flow in small diameter passageways. Magn. Reson. Med. 88 , (2022). Williams, S. D. et al. Neural activity induced by sensory stimulation can drive large-scale cerebrospinal fluid flow during wakefulness in humans. PLoS Biol. 21 , e3002035 (2023). Fultz, N. E. et al. Coupled electrophysiological, hemodynamic, and cerebrospinal fluid oscillations in human sleep. Science (80-. ). (2019) doi:10.1126/science.aax5440. Williamson, N. H., Komlosh, M. E., Benjamini, D. & Basser, P. J. Limits to flow detection in phase contrast MRI. J. Magn. Reson. Open 2 – 3 , (2020). Holstein-Rønsbo, S. et al. Glymphatic influx and clearance are accelerated by neurovascular coupling. Nat. Neurosci. 26 , 1042–1053 (2023). Charidimou, A. et al. The Boston criteria version 2.0 for cerebral amyloid angiopathy: a multicentre, retrospective, MRI–neuropathology diagnostic accuracy study. Lancet Neurol. 21 , (2022). Dreha-Kulaczewski, S. et al. Inspiration is the major regulator of human CSF flow. J. Neurosci. 35 , (2015). Munting, L. P. et al. Spontaneous vasomotion propagates along pial arterioles in the awake mouse brain like stimulus-evoked vascular reactivity. J. Cereb. Blood Flow Metab. (2023). Helakari, H. et al. Human NREM Sleep Promotes Brain-Wide Vasomotor and Respiratory Pulsations. J. Neurosci. 42 , (2022). Perosa, V. et al. Perivascular space dilation is associated with vascular amyloid-β accumulation in the overlying cortex. Acta Neuropathol. 143 , (2022). Eide, P. K. & Ringstad, G. Functional analysis of the human perivascular subarachnoid space. Nat. Commun. 15 , (2024). Harrison, I. F. et al. Non-invasive imaging of CSF-mediated brain clearance pathways via assessment of perivascular fluid movement with diffusion tensor MRI. Elife (2018) doi:10.7554/eLife.34028. Helenius, J. et al. Diffusion-weighted MR imaging in normal human brains in various age groups. Am. J. Neuroradiol. 23 , (2002). Bazin, P. L. et al. Sharpness in motion corrected quantitative imaging at 7T. Neuroimage 222 , (2020). Andersen, M., Björkman-Burtscher, I. M., Marsman, A., Petersen, E. T. & Boer, V. O. Improvement in diagnostic quality of structural and angiographic MRI of the brain using motion correction with interleaved, volumetric navigators. PLoS One 14 , (2019). Peper, E. S. et al. Highly accelerated 4D flow cardiovascular magnetic resonance using a pseudo-spiral Cartesian acquisition and compressed sensing reconstruction for carotid flow and wall shear stress. J. Cardiovasc. Magn. Reson. 22 , (2020). Uecker, M., Tamir, J. I., Ong, F. & Lustig, M. The BART Toolbox for Computational Magnetic Resonance Imaging. Ismrm (2016). Hirschler, L. et al. The driving force of glymphatics: influence of the cardiac cycle on CSF mobility in perivascular spaces in humans. in Proceedings of the 29th Annual Meeting of ISMRM, Sydney, Australia, 2020. Abstract 2127. Hirschler, L., Runderkamp, B., van Veluw, S. J., Caan, M. W. A. & van Osch, M. J. P. Effects of the cardiac and respiratory cycles on CSF-mobility in human subarachnoid and perivascular spaces. in Proceedings of the 31st Annual Meeting of ISMRM, London, UK, 2022. Abstract 0320. Klein, S., Staring, M., Murphy, K., Viergever, M. A. & Pluim, J. P. W. Elastix: a toolbox for intensity based medical image registration. IEEE Trans. Med. Imaging 29 , 196–205 (2010). Kroon, D.-J. DTI and Fiber Tracking. https://www.mathworks.com/matlabcentral/fileexchange/21130-dti-and-fiber-tracking (2008). Moreland, K. et al. The ParaView Guide. Sandia Natl. Lab. (2016). Jenkinson, M., Bannister, P., Brady, M. & Smith, S. Improved optimization for the robust and accurate linear registration and motion correction of brain images. Neuroimage 17 , (2002). Smith, S. M. Fast robust automated brain extraction. Hum. Brain Mapp. 17 , (2002). Beckmann, C. F. & Smith, S. M. Probabilistic Independent Component Analysis for Functional Magnetic Resonance Imaging. IEEE Trans. Med. Imaging 23 , (2004). Woolrich, M. W., Ripley, B. D., Brady, M. & Smith, S. M. Temporal autocorrelation in univariate linear modeling of FMRI data. Neuroimage 14 , (2001). Greve, D. N. & Fischl, B. Accurate and robust brain image alignment using boundary-based registration. Neuroimage 48 , (2009). Jenkinson, M. & Smith, S. A global optimisation method for robust affine registration of brain images. Med. Image Anal. 5 , (2001). Andersson, J. L. R., Jenkinson, M. & Smith, S. Non-linear registration aka spatial normalisation . FMRIB Technical Report TRO7JA2 (2007). Fischl, B. FreeSurfer. NeuroImage vol. 62 (2012). Zhang, Y., Brady, M. & Smith, S. Segmentation of brain MR images through a hidden Markov random field model and the expectation-maximization algorithm. IEEE Trans. Med. Imaging 20 , (2001). Yushkevich, P. A. et al. User-guided 3D active contour segmentation of anatomical structures: Significantly improved efficiency and reliability. Neuroimage 31 , 1116–1128 (2006). Manniesing, R. & Niessen, W. Multiscale vessel enhancing diffusion in CT angiography noise filtering. in Lecture Notes in Computer Science vol. 3565 (2005). Meijering, E. et al. Neurite tracing in fluorescence microscopy images using ridge filtering and graph searching: Principles and validation. in 2004 2nd IEEE International Symposium on Biomedical Imaging: Macro to Nano vol. 2 (2004). Team, J. JASP Team (Version 0.17). (2023). Hirschler, L. et al. High resolution T2-prepared MRI enables non-invasive assessment of CSF flow in perivascular spaces of the human brain. in Proceedings of the 28th Annual Meeting of ISMRM, Montréal, Canada, 2019. Abstract 0746. Van Osch, M. J. P., Petitclerc, L. & Hirschler, L. Probing cerebrospinal fluid mobility for human brain clearance imaging MRI: water transport across the blood-cerebrospinal fluid barrier and mobility of cerebrospinal fluid in perivascular spaces. Veins Lymphat. 11 , (2022). Additional Declarations Yes there is potential Competing Interest. The following conflicts of interest exist: The LUMC receives research support from Philips and Matthan Caan is a shareholder of Nicolab International Ltd. Supplementary Files Supplementaryinformation.docx CSF-STREAM supplementary information Videosglymphatics.pptx Supplementary Videos Videos.pptx CSF-STREAM videos Extendeddatafigures.docx Cite Share Download PDF Status: Published Journal Publication published 14 Oct, 2025 Read the published version in Nature Neuroscience → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3178346","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":342747403,"identity":"ae1687bc-93f0-460d-8ffa-7db12002e142","order_by":0,"name":"Lydiane Hirschler","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAwklEQVRIiWNgGAWjYBACAwYegwNAOoGPgYHxwQOwGHMDcVrYgEoNEsBijIS1MEC1sEkQq8XwcMUfuzw29t5nFYlt2+QZpBsJaeH/cPBsW3IxG89xsxuJbbcNG2QOEnbYwcaGA4ltEmlsIC0JDBKJRGhp+APUIv+MrYAELWwgW9jYGIjTwgxyWFtyYhtPGrNEwrnbhm2EtNi39xh/bPhjl9jPfozxw4ey2/L8EskH8GphYEYXYMOvfhSMglEwCkYBMQAAi+REaQlv9CcAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0003-2379-0861","institution":"C.J. Gorter MRI Center, Department of Radiology, Leiden University Medical Center, Leiden, The Netherlands","correspondingAuthor":true,"prefix":"","firstName":"Lydiane","middleName":"","lastName":"Hirschler","suffix":""},{"id":342747404,"identity":"e6bde597-7ada-48fe-a1ff-1d96c3f97f1e","order_by":1,"name":"Bobby A. Runderkamp","email":"","orcid":"https://orcid.org/0009-0007-5491-5379","institution":"Department of Radiology and Nuclear Medicine, Amsterdam University Medical Centers, University of Amsterdam, Amsterdam, The Netherlands","correspondingAuthor":false,"prefix":"","firstName":"Bobby","middleName":"A.","lastName":"Runderkamp","suffix":""},{"id":371901456,"identity":"0a2fa5a9-0cd9-443b-98db-9e49909d4788","order_by":2,"name":"Andreas Decker","email":"","orcid":"","institution":"German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany","correspondingAuthor":false,"prefix":"","firstName":"Andreas","middleName":"","lastName":"Decker","suffix":""},{"id":342747405,"identity":"7409e8e8-4710-441c-8780-79b9ae111f32","order_by":3,"name":"Thijs W. van Harten","email":"","orcid":"https://orcid.org/0000-0002-3408-9379","institution":"C.J. Gorter MRI Center, Department of Radiology, Leiden University Medical Center, Leiden, The Netherlands; Department of Neurology, Massachusetts General Hospital, Harvard Medical School, 175 Cambridge Street, Suite 300, Boston, MA 02114, USA","correspondingAuthor":false,"prefix":"","firstName":"Thijs","middleName":"W. van","lastName":"Harten","suffix":""},{"id":371901457,"identity":"a9461e93-b69d-4f7c-8ddd-69dcf9ad5695","order_by":4,"name":"Paul Scheyhing","email":"","orcid":"","institution":"German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany; Department of Neuroradiology, University Hospital Bonn, Bonn, Germany","correspondingAuthor":false,"prefix":"","firstName":"Paul","middleName":"","lastName":"Scheyhing","suffix":""},{"id":371901458,"identity":"b7f535bb-fb9e-41db-82e5-2856b2b3e59e","order_by":5,"name":"Philipp Ehses","email":"","orcid":"","institution":"German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany","correspondingAuthor":false,"prefix":"","firstName":"Philipp","middleName":"","lastName":"Ehses","suffix":""},{"id":342747406,"identity":"687d968e-cec4-4202-866f-b7893d2f180e","order_by":6,"name":"Léonie Petitclerc","email":"","orcid":"","institution":"C.J. Gorter MRI Center, Department of Radiology, Leiden University Medical Center, Leiden, The Netherlands","correspondingAuthor":false,"prefix":"","firstName":"Léonie","middleName":"","lastName":"Petitclerc","suffix":""},{"id":371901459,"identity":"32476609-ae5d-48d1-932c-f8d0e60c2717","order_by":7,"name":"Julia Nordsiek","email":"","orcid":"","institution":"German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany; Department of Vascular Neurology, University Hospital Bonn, Bonn, Germany","correspondingAuthor":false,"prefix":"","firstName":"Julia","middleName":"","lastName":"Nordsiek","suffix":""},{"id":371901460,"identity":"19ecd308-6253-439e-b9c7-a58bd09e7426","order_by":8,"name":"Eberhard Pracht","email":"","orcid":"","institution":"German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany","correspondingAuthor":false,"prefix":"","firstName":"Eberhard","middleName":"","lastName":"Pracht","suffix":""},{"id":342747407,"identity":"0f66319d-d12c-4642-a456-62bb828c9735","order_by":9,"name":"Bram F. Coolen","email":"","orcid":"","institution":"Department of Biomedical Engineering \u0026 Physics, Amsterdam University Medical Centers, Amsterdam, The Netherlands","correspondingAuthor":false,"prefix":"","firstName":"Bram","middleName":"F.","lastName":"Coolen","suffix":""},{"id":342747408,"identity":"1788ee17-a35b-481b-a89b-b0827704a074","order_by":10,"name":"Wietske van der Zwaag","email":"","orcid":"https://orcid.org/0000-0003-3223-9721","institution":"Spinoza Centre for Neuroimaging, Royal Netherlands Academy of Arts and Sciences, Amsterdam, The Netherlands; Computational Cognitive Neuroscience and Neuroimaging, Netherlands Institute for Neuroscience, KNAW, Amsterdam, The Netherlands","correspondingAuthor":false,"prefix":"","firstName":"Wietske","middleName":"van der","lastName":"Zwaag","suffix":""},{"id":371901461,"identity":"64907559-2dda-4c26-b216-570a2d5525b9","order_by":11,"name":"Tony Stöcker","email":"","orcid":"","institution":"German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany; Department of Physics and Astronomy, University of Bonn, Bonn, Germany","correspondingAuthor":false,"prefix":"","firstName":"Tony","middleName":"","lastName":"Stöcker","suffix":""},{"id":371901462,"identity":"71edb613-ae2b-47a6-81c2-36539ca76835","order_by":12,"name":"Philipp Vollmuth","email":"","orcid":"","institution":"Department of Neuroradiology, University Hospital Bonn, Bonn, Germany","correspondingAuthor":false,"prefix":"","firstName":"Philipp","middleName":"","lastName":"Vollmuth","suffix":""},{"id":371901463,"identity":"54df8671-9a5f-48bf-a4ff-f4954cd9437d","order_by":13,"name":"Daniel Paech","email":"","orcid":"","institution":"Department of Neuroradiology, University Hospital Bonn, Bonn, Germany; Department of Neuroradiology, University Hospital Bonn, Bonn, Germany; Department of Radiology, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA","correspondingAuthor":false,"prefix":"","firstName":"Daniel","middleName":"","lastName":"Paech","suffix":""},{"id":371901464,"identity":"0492baea-0298-40e8-9d19-80ece0d2544e","order_by":14,"name":"Alexander Effland","email":"","orcid":"","institution":"Institute of Applied Mathematics, University Bonn, Bonn, Germany","correspondingAuthor":false,"prefix":"","firstName":"Alexander","middleName":"","lastName":"Effland","suffix":""},{"id":342747409,"identity":"62cac878-f91e-4484-947f-392dc3a51e67","order_by":15,"name":"Marianne A.A. van Walderveen","email":"","orcid":"","institution":"C.J. Gorter MRI Center, Department of Radiology, Leiden University Medical Center, Leiden, The Netherlands","correspondingAuthor":false,"prefix":"","firstName":"Marianne","middleName":"A.A. van","lastName":"Walderveen","suffix":""},{"id":371901465,"identity":"e8a581f8-0452-453c-8bb1-da96e50e986c","order_by":16,"name":"Alexander Radbruch","email":"","orcid":"","institution":"German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany; Department of Neuroradiology, University Hospital Bonn, Bonn, Germany","correspondingAuthor":false,"prefix":"","firstName":"Alexander","middleName":"","lastName":"Radbruch","suffix":""},{"id":342747410,"identity":"f59cc02b-28f7-4a37-9b23-79a220e5c6a1","order_by":17,"name":"Mark A. van Buchem","email":"","orcid":"","institution":"C.J. Gorter MRI Center, Department of Radiology, Leiden University Medical Center, Leiden, The Netherlands","correspondingAuthor":false,"prefix":"","firstName":"Mark","middleName":"A. van","lastName":"Buchem","suffix":""},{"id":371901466,"identity":"49fa0d8b-20f6-4ab5-9e07-d90826dd30a4","order_by":18,"name":"Gabor C. Petzold","email":"","orcid":"","institution":"German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany; Department of Vascular Neurology, University Hospital Bonn, Bonn, Germany","correspondingAuthor":false,"prefix":"","firstName":"Gabor","middleName":"C.","lastName":"Petzold","suffix":""},{"id":342747411,"identity":"7822d013-0f57-4e3b-a624-89efc1697635","order_by":19,"name":"Susanne J. van Veluw","email":"","orcid":"","institution":"C.J. Gorter MRI Center, Department of Radiology, Leiden University Medical Center, Leiden, The Netherlands; Department of Neurology, Massachusetts General Hospital, Harvard Medical School, 175 Cambridge Street, Suite 300, Boston, MA 02114, USA","correspondingAuthor":false,"prefix":"","firstName":"Susanne","middleName":"J. van","lastName":"Veluw","suffix":""},{"id":342747412,"identity":"096dd450-dd2a-4960-b7ac-7e067d8d3f60","order_by":20,"name":"Matthan W.A. Caan","email":"","orcid":"","institution":"Department of Biomedical Engineering \u0026 Physics, Amsterdam University Medical Centers, Amsterdam, The Netherlands","correspondingAuthor":false,"prefix":"","firstName":"Matthan","middleName":"W.A.","lastName":"Caan","suffix":""},{"id":371901467,"identity":"5f6c30f4-9007-4924-8066-4814d0a67b13","order_by":21,"name":"Katerina Deike*","email":"","orcid":"","institution":"German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany; Department of Neuroradiology, University Hospital Bonn, Bonn, Germany","correspondingAuthor":false,"prefix":"","firstName":"Katerina","middleName":"","lastName":"Deike*","suffix":""},{"id":342747413,"identity":"a55da1be-6fd9-46df-b51a-db2beebe510c","order_by":22,"name":"Matthias J.P. van Osch*","email":"","orcid":"https://orcid.org/0000-0001-7034-8959","institution":"C.J. Gorter MRI Center, Department of Radiology, Leiden University Medical Center, Leiden, The Netherlands","correspondingAuthor":false,"prefix":"","firstName":"Matthias","middleName":"J.P. van","lastName":"Osch*","suffix":""}],"badges":[],"createdAt":"2023-07-17 14:25:37","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3178346/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3178346/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41593-025-02073-3","type":"published","date":"2025-10-14T04:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":66922810,"identity":"5cf2a09f-5516-47a0-858d-a992157f9b48","added_by":"auto","created_at":"2024-10-18 05:08:38","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":684297,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCSF-signal and CSF-mobility characteristics using CSF-STREAM\u003c/strong\u003e. CSF-signal, measured using the non-motion-sensitized reference scan, (A) in the subarachnoid space around the middle cerebral artery and (B) in perivascular spaces around penetrating arteries in one representative individual. Principal orientation of the CSF-mobility (C) in the subarachnoid space around the middle cerebral artery, including a zoomed area on one branch, and (D) in perivascular spaces of penetrating arteries. (C) and (D) are from the same regions of interest as (A) and (B). The line colors reflect the orientation of CSF-mobility: red colors indicate a left-right orientation, green an anterior-posterior orientation and blue a head-feet orientation. Volume rendering of (E) a CSF-mobility map (in mm\u003csup\u003e2\u003c/sup\u003e/s) and (F) a fractional anisotropy map in one representative individual.\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/4430e2323d78ff283b2d6b7b.png"},{"id":66921912,"identity":"f411b508-05e9-491b-948d-e93c3f6af32a","added_by":"auto","created_at":"2024-10-18 04:52:38","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":694310,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRegional CSF-mobility and fractional anisotropy using CSF-STREAM.\u003c/strong\u003e (A) Example of the location of the regions of interest (ROIs) in one representative individual: subarachnoid space (SAS) around the middle cerebral artery (MCA), 4\u003csup\u003eth\u003c/sup\u003e ventricle, CSF around the visual cortex, SAS of the motor cortex sulci, perivascular space (PVS) in the basal ganglia (BG), and PVS surrounding penetrating arteries. In each insert, the extracted volume rendering of the ROI is shown (top of the inserts) next to the CSF-mobility (in mm\u003csup\u003e2\u003c/sup\u003e/s) volume rendering within the ROI (bottom of the inserts). (B) Average CSF-mobility (in mm\u003csup\u003e2\u003c/sup\u003e/s) (left) and fractional anisotropy (FA) (right) in the different ROIs in eleven individuals. Each point represents the average value over voxels in each ROI per individual (one color per individual).\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/476b7e394cdc65b12e7efdcc.png"},{"id":66921917,"identity":"7ca1bb4a-18ca-4822-bee1-fbeed92ee039","added_by":"auto","created_at":"2024-10-18 04:52:38","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":601773,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCSF-mobility change across driving forces in large CSF-spaces around the circle of Willis.\u003c/strong\u003e (A) Maps of CSF-mobility change from the mean value over phases (in %) across the cardiac (top), respiration (middle), and random (bottom) cycles in one representative individual. (B) Voxel-wise CSF-mobility changes in three regions of interest shown in the insert. Each colored line represents the signal of an individual voxel within the ROI, the thicker line the mean value over voxels in that ROIand the shaded area the confidence interval over voxels.\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/da7165a16809e59d4f689cf2.png"},{"id":66924089,"identity":"5717bd97-2e2e-42c6-8933-9fbd9094cac0","added_by":"auto","created_at":"2024-10-18 05:24:39","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":406656,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCSF-mobility change across driving forces in PVS around penetrating arteries.\u003c/strong\u003e (A) Maps of CSF-mobility change from the mean value over phases (in %) across the cardiac (top), respiration (middle), and random (bottom) cycles in one representative individual. (B) Voxel-wise CSF-mobility changes in three regions of interest shown in the insert. Each colored line represents the signal of an individual voxel within the ROI, the thick line the mean value over voxels in that ROI and the shaded area the confidence interval over voxels.\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/1e3143cb0bad62659b939b0d.png"},{"id":66921921,"identity":"cff1b766-05b4-4bdf-909b-29c91ac9dea2","added_by":"auto","created_at":"2024-10-18 04:52:39","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":325557,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRegional comparison of CSF-mobility change across driving forces.\u003c/strong\u003e CSF-mobility change from the mean value over phases (in %) across (A) the cardiac cycle (pink), (B) respiratory cycle (green) and (A\u0026amp;B) random cycle (grey) in six regions of interest in eleven individuals. Note that the y-axis range is different for the 4\u003csup\u003eth\u003c/sup\u003e ventricle region of interest. Each line represents the mean over individuals and the shaded error areas represent confidence intervals of SD×1.96/√n (n=11).\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/774423ee9ad15c434e3b3205.png"},{"id":66922626,"identity":"7c960b7d-43bc-47e4-bf44-5d1623d0706b","added_by":"auto","created_at":"2024-10-18 05:00:39","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":484314,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRegional comparison of CSF-mobility fit quality and amplitude change across driving forces.\u003c/strong\u003e Comparison between the effect of the cardiac, respiration and random binning on (A) the fit quality and (B) the amplitude of the CSF-mobility change (%), in eleven individuals. Note that the y-axis range is different for the 4\u003csup\u003eth\u003c/sup\u003e ventricle and SAS sulci regions of interest in panel B. Each datapoint represents the value per individual in a region of interest. Asterisks indicate significant differences with a p-value\u0026lt;0.01 using a Wilcoxon signed rank test applied when Friedman’s test was significant.\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/7bfcad759d9ed7b80c5a8e3d.png"},{"id":66922815,"identity":"45aa7bd0-8781-4327-9c34-3d653b4169d9","added_by":"auto","created_at":"2024-10-18 05:08:39","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":356670,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEffects of entrained vasomotion on CSF-mobility and comparison to other driving forces. \u003c/strong\u003e(A) Average ± (SD×1.96/√n) CSF-mobility (in mm\u003csup\u003e2\u003c/sup\u003e/s) at rest (black) and in presence (grey) of a 0.1 Hz visual stimulation to entrain vasomotion in nine individuals. Each color represents the value in one individual. Asterisks indicates significant (p=0.004) changes between the two conditions using a Wilcoxon signed rank test. (B) Change in CSF-mobility (in %) induced by the 0.1 Hz visual stimulation compared to rest in the visual cortex (defined as the region where the BOLD z-score was higher than 7) and in a control region (where the BOLD z-score was lower than 1), in nine individuals. (C) CSF-mobility change (in %) in the visual cortex versus BOLD amplitude change (in %) in nine individuals. (D) Change in CSF-mobility (in %) induced by different driving forces (cardiac, respiration, visual stimulation and random) in six individuals who participated in both studies.\u003c/p\u003e","description":"","filename":"image7.png","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/d57c545ac6abe2fc43004c89.png"},{"id":66921918,"identity":"75fad5a6-3b8e-4419-84d2-53f69f4ca089","added_by":"auto","created_at":"2024-10-18 04:52:38","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":432807,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCSF-STREAM in CAA patients versus healthy controls.\u003c/strong\u003e (A) Example of a 1-mm CSF-rim in the subarachnoid space (SAS) around the middle cerebral artery (MCA). (B) CSF-mobility is significantly increased (p=0.01, Mann-Whitney U-test) and (C) FA is significantly decreased (p=0.02, Mann-Whitney U-test) in the 1 mm-thick SAS around the MCA of CAA patients (pink) versus healthy controls (black). (D) ROI volume around the MCA in controls and CAA patients. (E) Example of perivascular space (PVS) segmentation around penetrating vessels in the centrum semiovale (CSO). No significant change in (F) CSF-mobility nor (G) FA was found in PVS. (H) The PVS volume was significantly increased (p = 0.007, Mann-Whitney U-test) in CAA patients. Each datapoint represents the value per individual in a region of interest (ROI).\u003c/p\u003e","description":"","filename":"image8.png","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/1bc0866a12cc34a7b98f89b1.png"},{"id":93558495,"identity":"15b8d954-f7de-4203-8894-549d87055ac7","added_by":"auto","created_at":"2025-10-15 07:06:48","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5111246,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/998ce277-8803-45ac-979c-096e20545e9f.pdf"},{"id":66922621,"identity":"e6aee3b8-e6a8-4316-a47c-b6d7804b8ab6","added_by":"auto","created_at":"2024-10-18 05:00:38","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":138477,"visible":true,"origin":"","legend":"\u003cp\u003eCSF-STREAM supplementary information\u003c/p\u003e","description":"","filename":"Supplementaryinformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/d2174db9542b003f76dce894.docx"},{"id":66921924,"identity":"fae13ae1-3e4c-47a3-84cc-a4df1e9f0911","added_by":"auto","created_at":"2024-10-18 04:52:39","extension":"pptx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":37068769,"visible":true,"origin":"","legend":"Supplementary Videos","description":"","filename":"Videosglymphatics.pptx","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/7bcdd458251e812256eb2cee.pptx"},{"id":66921923,"identity":"0899fd5e-930d-4f0c-894c-440fa606ec85","added_by":"auto","created_at":"2024-10-18 04:52:39","extension":"pptx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":37063647,"visible":true,"origin":"","legend":"CSF-STREAM videos","description":"","filename":"Videos.pptx","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/b3729039ef5e6be9a7d8bd07.pptx"},{"id":66922624,"identity":"87967d30-1033-42a2-9e23-8a6e471616fc","added_by":"auto","created_at":"2024-10-18 05:00:39","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":1139959,"visible":true,"origin":"","legend":"","description":"","filename":"Extendeddatafigures.docx","url":"https://assets-eu.researchsquare.com/files/rs-3178346/v1/3a1e792989c3fa7fc759f314.docx"}],"financialInterests":"\u003cb\u003eYes\u003c/b\u003e there is potential Competing Interest.\nThe following conflicts of interest exist: The LUMC receives research support from Philips and Matthan Caan is a shareholder of Nicolab International Ltd.","formattedTitle":"Region specific drivers of cerebrospinal fluid mobility as measured by high-resolution non-invasive MRI in humans","fulltext":[{"header":"Highlights","content":"\u003cul\u003e\n \u003cli\u003eCSF-mobility can be measured in humans down to the level of perivascular spaces surrounding penetrating\u0026nbsp;vessels using CSF-STREAM\u003c/li\u003e\n \u003cli\u003eIn perivascular spaces surrounding penetrating\u0026nbsp;vessels, the cardiac and respiratory cycles have similar effects on CSF-mobility\u003c/li\u003e\n \u003cli\u003eEntraining vasomotion at 0.1 Hz can drive CSF-mobility in humans\u003c/li\u003e\n \u003cli\u003eCSF-mobility is altered in patients with cerebral amyloid angiopathy\u003c/li\u003e\n\u003c/ul\u003e"},{"header":"Introduction","content":"\u003cp\u003eDue to its high metabolic rate, the brain produces large quantities of proteins, whose abnormal accumulation is involved in a number of pathologies and neurodegenerative disorders.\u0026nbsp;However, unlike other organs in the body, the brain\u0026nbsp;tissue\u0026nbsp;lacks a classic lymphatic system\u0026nbsp;to transport excess soluble proteins it produces (\u0026lsquo;waste\u0026rsquo;) out of the brain (e.g. to\u0026nbsp;lymph nodes\u0026nbsp;or\u0026nbsp;dural lymphatic\u0026nbsp;vessels).\u0026nbsp;The\u0026nbsp;true nature of brain clearance mechanisms has eluded characterization for centuries, until the recent uptick in interest in the topic\u003csup\u003e1,2\u003c/sup\u003e. The use of microscopy in rodents\u003csup\u003e3\u0026ndash;5\u003c/sup\u003e and invasive intrathecal injections in humans\u003csup\u003e6,7\u003c/sup\u003e has allowed researchers to unravel and describe many aspects of new brain clearance pathways and more subtle physiological processes, and their results have opened up new questions and debate. While we know that brain clearance processes must involve cerebrospinal fluid (CSF) as the main carrier of \u0026lsquo;waste\u0026lsquo; products, likely along pathways surrounding small blood vessels called perivascular spaces (PVS)\u003csup\u003e8\u003c/sup\u003e, the exact anatomical pathway(s), driving force(s) as well as physiological processes involved in CSF-mediated brain clearance remain unresolved\u003csup\u003e1,9,10\u003c/sup\u003e. Some studies suggest the presence of an active mechanism driving CSF-flow along PVS (including the glymphatic\u003csup\u003e3\u003c/sup\u003e and intramural periarterial drainage\u003csup\u003e11\u003c/sup\u003e theories), whereas others propose that perivascular clearance mainly occurs via more passive mixing mechanisms\u003csup\u003e9,12\u003c/sup\u003e. In all proposed mechanisms, CSF-mobility in PVS would be facilitated by physiological motion processes, such as cardiac pulsations\u003csup\u003e4\u003c/sup\u003e, respiration\u003csup\u003e13\u003c/sup\u003e, or vasomotion\u003csup\u003e5\u003c/sup\u003e. As such, it has been suggested that these drivers of motion can propel soluble \u0026lsquo;waste\u0026rsquo; products from the PVS up towards the pial surface, where bulk flow assures further egress.\u003c/p\u003e\n\u003cp\u003eStudying\u0026nbsp;CSF-mediated\u0026nbsp;brain clearance and its driving forces is of particular importance since clearance failure\u0026nbsp;has been implicated in the\u0026nbsp;accumulation of toxic proteins in the brain, such as amyloid-\u0026beta;\u0026nbsp;and tau. A\u0026nbsp;plethora of neurological diseases like Alzheimer\u0026rsquo;s disease\u003csup\u003e14\u003c/sup\u003e, cerebral amyloid angiopathy (CAA)\u003csup\u003e15\u003c/sup\u003e, traumatic brain injury\u003csup\u003e16\u003c/sup\u003e, and ischemic stroke\u003csup\u003e17\u003c/sup\u003e are associated with brain clearance deficiencies. CAA, especially, is a common small vessel disease and leading cause of hemorrhagic stroke and dementia in older individuals, characterized by the accumulation of amyloid-\u0026beta;\u0026nbsp;in the vessel wall, possibly due to impaired CSF-mediated amyloid-\u0026beta;\u0026nbsp;clearance\u003csup\u003e15,18\u003c/sup\u003e. CAA frequently co-occurs with Alzheimer\u0026rsquo;s pathology and is associated with increased risk of developing amyloid related imaging abnormalities in the context of anti-amyloid immunotherapy\u003csup\u003e15\u003c/sup\u003e. Notably, in a rat model of CAA, the mobility of CSF in the subarachnoid space (SAS) surrounding large arteries was recently found to be increased\u003csup\u003e19\u003c/sup\u003e. This was accompanied by a reduction of volume of tracer transport to the brain tissue, altogether suggesting that CSF would bypass the brain tissue due to amyloid-\u0026beta;\u0026nbsp;deposits. However, it is currently unknown whether these recent findings in rodents also translate to humans with CAA. A better understanding of\u0026nbsp;CSF-mediated\u0026nbsp;brain clearance and its driving forces is therefore urgent, as it would provide crucial new avenues not only to elucidate the pathophysiology of these neurological diseases, but also towards new targets for efficient therapeutic strategies to slow or stop disease progression.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eUnfortunately, current knowledge of CSF-mediated brain clearance mechanisms is mostly derived from experimental studies performed in rodents, associated with significant limitations. First, these studies tend to use techniques which introduce perturbations to the very system they aim to characterize, such as euthanasia prior to measurement, which may result in the collapse of essential structures for brain clearance\u003csup\u003e4\u003c/sup\u003e, anesthesia, which interferes with hemodynamics and likely with brain clearance as well, or invasive imaging techniques\u003csup\u003e3\u0026ndash;5,20\u003c/sup\u003e like cranial windows and\u0026nbsp;injection of fluorescent dyes, which may induce local pressure changes and thereby affect physiological CSF motion. Second, it remains unclear how experimental findings in rodents translate to humans given the inherent differences in brain size and physiological parameters between species\u003csup\u003e2\u003c/sup\u003e. For example, cardiac frequency is about seven to twelve times slower in humans compared to mice\u003csup\u003e21\u003c/sup\u003e, whereas vasomotion frequency is comparable, centered around 0.1 Hz. This could\u0026nbsp;influence the relative contributions to the suggested driving forces of CSF flow, i.e. the cardiac cycle, respiration, and vasomotion. Moreover, to the best of our knowledge, current in vivo studies only focus on motion of CSF in the SAS and in the ventricular system\u003csup\u003e22\u0026ndash;26\u003c/sup\u003e, lacking the spatial resolution to investigate CSF-mobility in PVS around penetrating vessels, which are believed to be the channels along which soluble \u0026lsquo;waste\u0026rsquo; products clear out of the brain.\u003c/p\u003e\n\u003cp\u003eA non-invasive method that measures CSF-mobility at a high spatial resolution is needed to further understand perivascular brain clearance mechanisms in humans, allowing the assessment of CSF-mobility directly where clearance is thought to occur, i.e. within CSF-filled PVS. This would pave the way for studying\u0026nbsp;CSF-mediated brain\u0026nbsp;clearance in larger sample sizes, patient cohorts, and in longitudinal follow-up studies.\u0026nbsp;Magnetic resonance imaging (MRI) at a high magnetic field strength (7 Tesla) is an excellent modality for such a non-invasive imaging strategy, as it provides the necessary resolution to image the PVS and has the added benefit of easily differentiating CSF from other brain tissues or fluids by exploiting its specific magnetic properties.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn this study, we present a non-invasive, high-resolution, and CSF-specific MRI technique that allows the characterization of CSF-mobility, even in PVS around penetrating arteries. We also investigate and compare the influence of the cardiac cycle, respiratory cycle, and vasomotion as driving forces for CSF-mobility. Lastly, we apply the newly developed technique in patients with CAA in a pilot study, in order to explore the potential of this technique to improve our understanding of neurodegenerative diseases.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cem\u003eHigh-resolution imaging of CSF-mobility is achieved using CSF-STREAM \u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eWhole-brain CSF-signal was visualized in twenty healthy, younger individuals (age 33\u0026plusmn;12 years, 16 females, 4 males) at rest using ultra-high field (7 Tesla) MRI with a T\u003csub\u003e2\u003c/sub\u003e-prepared high-resolution (i.e. 0.45 mm isotropic voxel-size) accelerated readout with a long echo-time (Fig. 1A-B, Fig. 2A, Extended data Fig. 1 and Suppl. video 1). Importantly, signal originating from blood and brain tissue was suppressed (i.e. not significantly different from the noise level), such that the CSF-signal was successfully isolated (Extended data Fig. 1). The introduction of motion-sensitizing gradients of 3.5\u0026nbsp;mm/s in the T\u003csub\u003e2\u003c/sub\u003e-preparation module, which during repeated measurements encode mobility in six orthogonal directions, allowed calculation of a tensor from which the CSF-mobility, fractional anisotropy (FA) and principal CSF-mobility orientation were computed (Fig. 1). CSF-mobility is measured and calculated in a similar way as an apparent diffusion coefficient acquired at a very low b-value, making the MRI sequence more sensitive to flow than diffusion\u003csup\u003e27\u003c/sup\u003e. The term \u0026lsquo;mobility\u0026rsquo; is used as opposed to \u0026lsquo;flow\u0026rsquo; or \u0026lsquo;diffusion\u0026rsquo; to accentuate that the dephasing underlying the signal attenuation is caused by either slow flow, laminar flow or by back-and-forth motion of CSF, or a combination of all processes\u003csup\u003e22,27\u003c/sup\u003e. Altogether, the proposed CSF-Selective T\u003csub\u003e2\u003c/sub\u003e-prepared REadout with Acceleration and Mobility-encoding (CSF-STREAM) provides a fully non-invasive technique to quantitatively measure the mobility of CSF at an unprecedented high resolution: from the ventricles to the SAS around large vessels, down to small PVS in the basal ganglia and around penetrating arteries (Fig. 2 and Extended data Fig. 2).\u003c/p\u003e\n\u003cp\u003eIn the SAS around the MCA and in PVS, CSF mainly moves along the vessel as represented by the principal vector orientation (Fig. 1C-D). The FA in PVS (Fig. 2B) was high, further substantiating that CSF preferentially moves along one orientation. CSF-mobility was higher at the base of the brain than at the brain surface (Fig. 1E, Fig. 2B). It was highest in the SAS around the middle cerebral artery (0.041\u0026plusmn;0.008 mm\u003csup\u003e2\u003c/sup\u003e/s) and three times lower in PVS of the basal ganglia and of penetrating arteries (0.012\u0026plusmn;0.003 mm\u003csup\u003e2\u003c/sup\u003e/s and 0.015\u0026plusmn;0.005 mm\u003csup\u003e2\u003c/sup\u003e/s, respectively).\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eRegion-\u003c/em\u003e\u003cem\u003especific effects of cardiac and respiratory cycles on CSF-mobility\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eNext, the influence of driving forces of CSF-mobility was assessed in eleven individuals by studying the effect of the cardiac and respiratory cycles on CSF-mobility, as these are proposed to be driving forces of CSF-mobility\u003csup\u003e2\u003c/sup\u003e. After retrospective binning of k-space to the recordings of external physiological sensors (Extended data Fig. 3), we observed that CSF-mobility and FA were indeed fluctuating across the cardiac and respiratory cycles in an oscillatory manner (Fig. 3, Fig. 4). We found that in selected large CSF-spaces located at the base of the brain, i.e. the SAS around the MCA (SAS-MCA) and the 4\u003csup\u003eth\u003c/sup\u003e ventricle, the cardiac cycle was associated with significantly larger CSF-mobility oscillations than respiration. This is shown by the average CSF-mobility changes across phases in Fig. 5 and the significantly better fit quality to a sinusoid (SAS-MCA: 0.7\u0026plusmn;0.06 for cardiac versus 0.5\u0026plusmn;0.08 for respiration, p=0.002; 4\u003csup\u003eth\u003c/sup\u003e ventricle: 0.8\u0026plusmn;0.08 for cardiac versus 0.6\u0026plusmn;0.1 for respiration, p=0.002; Wilcoxon signed rank tests) and CSF-mobility change amplitude (SAS-MCA: 3.2\u0026plusmn;0.6 % for cardiac versus 1.2\u0026plusmn;0.4 % for respiration, p\u0026lt;0.001; 4\u003csup\u003eth\u003c/sup\u003e ventricle: 8.4\u0026plusmn;2.9 % for cardiac versus 2.5\u0026plusmn;1.0 % for respiration, p\u0026lt;0.001; Wilcoxon signed rank tests), as shown in Fig. 6. In contrast, in the smaller PVS in the basal ganglia and around penetrating arteries, CSF-mobility was driven by similar contributions of the cardiac and respiratory cycles (Fig. 5 and Fig. 6): the cardiac cycle induced a CSF-mobility change amplitude of 2.5\u0026plusmn;0.5 % in basal ganglia PVS and 2.8\u0026plusmn;0.7 % in PVS around penetrating arteries, whereas respiration led to 2.4\u0026plusmn;0.7 % (basal ganglia PVS) and 2.6\u0026plusmn;0.6 % (penetrating PVS) CSF-mobility change amplitudes.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOur method of retrospective binning did not artificially generate the observed CSF-mobility changes, as random binning did not result in a clear oscillatory pattern across phases (Fig. 5): CSF-mobility changes across random phases in a zig-zag-like fashion with e.g. several dips and peaks within one cycle, indicating a noisier pattern than that observed across cardiac and respiration phases. This can also be concluded quantitatively from the fit quality, fit amplitude, and the amount of voxels with coherent changes across phases (i.e. voxels with R\u003csup\u003e2\u003c/sup\u003e\u0026gt;0.5), which were all significantly lower for random ordering compared to binning based on cardiac or respiratory traces (Fig. 6, Extended data Fig. 5). Moreover, spatially coherent patterns (e.g. a right-left symmetry) in CSF-mobility changes can be observed for cardiac and respiration binning but not for random ordering (Fig. 3 and Suppl. video 2), providing further confidence that these oscillations are indeed originating from physiological pulsations and are not an artificial result of the reconstruction step.\u003c/p\u003e\n\u003cp\u003eAlong with the cardiac and respiratory cycles, vasomotion has been proposed as one of the driving forces for CSF. Unfortunately, there is no external physiological sensor to detect vasomotion that could be used for retrospective binning. A previous study in rodents showed that a 0.1\u0026nbsp;Hz visual stimulation could entrain vasomotion and lead to faster clearance of fluorescent tracers in the mouse brain\u003csup\u003e5\u003c/sup\u003e. Moreover, a recent study showed that visual stimulation can enhance CSF-flow in the 4\u003csup\u003eth\u003c/sup\u003e ventricle\u003csup\u003e25\u003c/sup\u003e. To investigate whether visual stimulation also drives local CSF-mobility in the SAS and PVS of the visual area in humans, a flashing checkerboard stimulus was shown to nine individuals while measuring CSF-mobility. Entraining vasomotion by means of 0.1\u0026nbsp;Hz visual stimulation significantly increased CSF-mobility in the visual cortex by 1.2 % on average (range = 0.2 - 3.4 %, Wilcoxon signed rank test p=0.004, Fig. 7A-B), compared to a control region outside of the stimulated area (average difference in control region = 0.01 %, range = -0.35 - 0.36 %, n=9, Wilcoxon signed rank test p=0.82). Note that the \u0026ldquo;visual cortex\u0026rdquo; region consists of the CSF within the activated area in response to the visual stimulation (see methods). To investigate whether the spread in increase in CSF-mobility across individuals originated from different responses to visual stimulation, we calculated the effect of visual stimulation on evoked vascular reactivity by quantifying the BOLD signal amplitude change. A Bayesian correlation analysis gave anecdotal evidence for a positive correlation between BOLD amplitude and CSF-mobility change (BF\u003csub\u003e0+\u003c/sub\u003e=2.3, Fig. 7C), suggesting that the vessel wall displacement induced by visual stimulation potentially leads to the increase in CSF-mobility\u003csup\u003e28\u003c/sup\u003e. The CSF-mobility increase induced by entraining vasomotion approaches the amplitude change driven by cardiac and respiratory cycles (Fig. 7D).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eRegional alterations of CSF-mobility and FA in patients with CAA\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eLastly, to show the potential of CSF-STREAM to improve our understanding of neurodegenerative diseases, we applied the MRI-sequence in eight patients with a clinical diagnosis of sporadic CAA, and in eight age- and sex-matched healthy controls.\u003c/p\u003e\n\u003cp\u003eWe found a 20% increase in CSF-mobility in the SAS closely surrounding the MCA of CAA patients as compared to healthy controls (CSF-mobility\u003csub\u003eCAA\u003c/sub\u003e = 0.042\u0026plusmn;0.006\u0026nbsp;mm\u003csup\u003e2\u003c/sup\u003e/s, CSF-mobility\u003csub\u003eControl\u003c/sub\u003e = 0.035\u0026plusmn;0.005\u0026nbsp;mm\u003csup\u003e2\u003c/sup\u003e/s, p=0.01, Mann-Whitney U-test, Fig. 8B). This increase in CSF-mobility was accompanied with a 10% decrease in FA (FA\u003csub\u003eCAA\u003c/sub\u003e = 0.65\u0026plusmn;0.06\u0026nbsp;mm\u003csup\u003e2\u003c/sup\u003e/s, FA\u003csub\u003eControl\u003c/sub\u003e = 0.73\u0026plusmn;0.04\u0026nbsp;mm\u003csup\u003e2\u003c/sup\u003e/s, p=0.02, Mann-Whitney U-test, Fig. 8C). Of note, this significant group difference was detectable in the SAS around the MCA up to a radius of approximately 1.7\u0026nbsp;mm around the vessel; beyond this distance, the difference diminished (Extended data Fig. 7).\u003c/p\u003e\n\u003cp\u003eIn PVS around penetrating vessels in the centrum semiovale (CSO), CSF-mobility and FA were not significantly different between groups (CSF-mobility\u003csub\u003eCAA \u003c/sub\u003e= 0.016\u0026plusmn;0.006\u0026nbsp;mm\u0026sup2;/s, CSF-mobility\u003csub\u003eControl \u003c/sub\u003e= 0.017\u0026plusmn;0.006\u0026nbsp;mm\u0026sup2;/s, p \u0026ge; 0.05, Mann-Whitney U-test), whereas a significantly increased PVS volume was found in CAA patients compared to controls, as expected\u003csup\u003e29\u003c/sup\u003e (Fig. 8H, p = 0.007, Mann-Whitney U-test). Age, sex, lifestyle factors as well as cognitive and physical health scores did not differ between both groups (Extended data Fig. 8).\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study we measured and directly compared the effects induced by various driving forces not only in large CSF-spaces such as the 4\u003csup\u003eth\u003c/sup\u003e ventricle or in the SAS at the brain surface, as was shown by previous studies\u003csup\u003e4,24\u0026ndash;26,30\u003c/sup\u003e, but also for the first time in smaller PVS around penetrating arteries, which are closer to waste production sites. Moreover, we provide the first in vivo evidence of altered CSF-mobility in patients with CAA, a disease associated with brain clearance impairment.\u003c/p\u003e\n\u003cp\u003eA strength of our results in the younger, healthy cohort is that a direct comparison can be made between the cardiac/respiration/random cycles as they all originate from the same dataset, i.e. one dataset was reconstructed in different ways, thus capturing the individual in the same physiological state. Our study shows that driving forces for CSF-mobility are region specific. In large CSF-spaces located at the base of the brain (SAS around the MCA and 4\u003csup\u003eth\u003c/sup\u003e ventricle), the cardiac cycle appears to be a larger contributing driving force to CSF-mobility as compared to respiration. This is in line with previous findings in mice\u003csup\u003e4\u003c/sup\u003e and humans\u003csup\u003e23,30\u003c/sup\u003e and can be explained by the fact that during systole the brain expands due to increased filling of its entire arterial system, which leads to CSF flowing out of the skull to compensate for the increased brain volume. This flow reverses during diastole. Conversely, in PVS, located around penetrating arteries more distally in the vascular tree where cardiac pulsations are likely dampened down, our results point to similar influence on CSF-mobility fluctuations by cardiac and respiratory cycles. This new finding in PVS suggests both cardiac and respiratory rhythms are equally important driving forces for perivascular CSF-mobility. This would support the notion of multiple sources promoting mixing within PVS\u003csup\u003e12\u003c/sup\u003e, as opposed to a single wave caused by a single physiological phenomenon (e.g. only arterial pulsations) pushing CSF through PVS\u003csup\u003e4,11\u003c/sup\u003e. It should, however, be understood that in our study the six cardiac and respiratory phases originate from data collected over 40 minutes. Therefore, possible effects from different types of respiration (e.g. single events of deep breaths) would be averaged out. It can be hypothesized that forced breathing or deep inspirations could further enhance CSF-mobility fluctuations, as shown previously\u003csup\u003e24,30\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLow-frequency (~0.1\u0026nbsp;Hz) vasomotion has also been proposed as a driving force for brain clearance, and was previously experimentally shown to be associated with increased tracer movement in CSF\u003csup\u003e5,11\u003c/sup\u003e. The influence of resting-state vasomotion can, unfortunately, not be assessed by retrospective binning as it was done with the cardiac and respiratory cycles, since it does not provide an accessible external triggering opportunity. Animal experiments have shown that low frequency visual stimulation can be used to entrain vasodilation at the vasomotion frequency and thereby speed up clearance of fluorescent-labeled tracers\u003csup\u003e5\u003c/sup\u003e. Moreover, in humans it was recently shown that a visual stimulation could drive global CSF-flow\u0026nbsp;in\u0026nbsp;the 4\u003csup\u003eth\u003c/sup\u003e ventricle\u003csup\u003e25\u003c/sup\u003e.\u0026nbsp;By using a similar stimulation strategy to entrain vasomotion, we showed that CSF-mobility within the visual cortex area\u0026nbsp;itself\u0026nbsp;is significantly enhanced in humans during 0.1 Hz frequency visual stimulation and that the amplitude of this effect approaches the strength of cardiac and respiratory fluctuations. These findings might even lead to the provocative hypothesis that, similar as in rodents, a simple visual stimulation paradigm at an optimal frequency may be able to enhance brain clearance in the visual cortex\u003csup\u003e31\u003c/sup\u003e. The ability\u0026nbsp;of CSF-STREAM\u0026nbsp;to measure CSF-mobility\u0026nbsp;within the stimulated region, i.e.\u0026nbsp;much closer to the production sites of metabolic \u0026lsquo;waste\u0026rsquo; products, should be considered an important advantage over techniques that measure CSF-flow at a single location at the exit of the brain, i.e. the aqueduct or the 4\u003csup\u003eth\u003c/sup\u003e ventricle.\u003c/p\u003e\n\u003cp\u003eThe observed increase in CSF-mobility by entraining vasomotion might mirror similar, albeit stronger, physiological fluctuations which occur during sleep. Indeed, in sleep, low frequency oscillations in cerebral blood volume, measured using BOLD fMRI\u003csup\u003e26,32\u003c/sup\u003e, are of similar amplitude as the ones measured in our study during a 0.1 Hz visual stimulation (Fig. 7C). As these oscillations are brain-wide during sleep, and not localized to a stimulated region, this could in turn lead to brain-wide enhancement of CSF-mobility during sleep.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWhile the increase in CSF-mediated clearance during sleep is proposed to be a key feature of its proper function, clearance failure has been suggested to be part of the pathogenesis of CAA, a common small vessel disease and contributor to dementia, characterized by vascular amyloid-\u0026beta;\u0026nbsp;accumulation and subsequent brain lesions such as hemorrhages. Our pilot study in CAA patients provides the first in vivo confirmation of observations made in rodent models of CAA. We observed regional alterations in CSF-mobility and FA in patients with sporadic CAA compared to healthy controls, even in our relatively small sample size: CSF-mobility was found to be increased by 20% and FA decreased by 10% within the SAS surrounding the MCA, whereas these parameters remained unchanged in PVS. Our findings are in line with the observations in the rodent model of CAA, where a similar 20% increase in CSF-speed was found in the SAS at the skull base, whereas CSF-speed in tissue remained unchanged\u003csup\u003e19\u003c/sup\u003e. The higher CSF-mobility together with lower FA might indicate higher but more disorganized CSF-mobility patterns\u0026nbsp;in the SAS\u0026nbsp;with CSF bypassing the tissue compartment. Such bypassing might be caused directly by amyloid-\u0026beta;\u0026nbsp;accumulation in cortical or leptomeningeal vessels or indirectly by stiffening of arterioles and loss of smooth muscle cells\u003csup\u003e33\u003c/sup\u003e. Interestingly, we found that the distance\u0026nbsp;of the CSF-volume\u0026nbsp;to the vessel is of importance in the SAS surrounding the MCA (Extended\u0026nbsp;data\u0026nbsp;Fig.\u0026nbsp;7): when measuring\u0026nbsp;in CSF located\u0026nbsp;within ~1.7 mm from the MCA, CSF-mobility and FA were found different between CAA and healthy controls, whereas when including CSF located further away from the MCA, the changes between groups attenuated. This might be explained by an additional membrane in the SAS along the major cerebral arteries that compartmentalizes the SAS,\u0026nbsp;as recently proposed by Eide and Ringstad\u003csup\u003e34\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe newly introduced CSF-STREAM technique was inspired by the long echo-time diffusion tensor imaging approach\u0026nbsp;(DTI)\u0026nbsp;proposed by Harrison and colleagues to measure CSF-mobility in the SAS surrounding the MCA in the rat brain\u003csup\u003e35\u003c/sup\u003e. The isolation of CSF-signal is essential to the successful measurement of CSF-mobility in PVS, as it ensures that the measured signal solely originates from CSF and not from slow-flowing blood in the vessel located inside of the PVS. A good separation of the CSF-signal was achieved, as shown in Fig. 1, Extended data Fig. 1 and Suppl. video 1. The use of motion-sensitizing gradients applied during T\u003csub\u003e2\u003c/sub\u003e-preparation instead of a traditional diffusion sequence enabled the robust use of a multi-shot turbo-spin-echo (TSE) readout for which the image quality and the achievable resolution is significantly higher than with an EPI readout that is usually the backbone of DTI-scans: the motion information encoded in the T\u003csub\u003e2\u003c/sub\u003e-preparation is stored in the longitudinal plane and therefore avoids phase accruals that would make the multi-shot reconstruction subject to image artefacts. Altogether, using the proposed CSF-STREAM technique, we showed that CSF-mobility could be measured non-invasively and at a high spatial resolution in healthy individuals and patients by combining accelerated ultra-high-field MRI while exploiting the magnetic properties of CSF and motion-sensitizing gradients.\u003c/p\u003e\n\u003cp\u003eThe measured CSF-mobility values are more than ten times higher than water diffusion values, even in PVS (~1\u0026times;10\u003csup\u003e-3\u003c/sup\u003emm\u003csup\u003e2\u003c/sup\u003e/s for water diffusion in the brain tissue\u003csup\u003e36\u003c/sup\u003e versus\u0026nbsp;10-50\u0026times;10\u003csup\u003e-3\u003c/sup\u003emm\u003csup\u003e2\u003c/sup\u003e/s for CSF-mobility). This high CSF-mobility combined with a high FA indicates that the physiological process behind CSF-mobility is not pure diffusion or plug flow, but more likely laminar flow or back-and-forth motion\u003csup\u003e22\u003c/sup\u003e. This is further substantiated by the influence of physiology on CSF-mobility: if the measured CSF-mobility was diffusion-dominant, then no effect of the cardiac or respiratory cycles should have been observed\u003csup\u003e22,23,27\u003c/sup\u003e. The motion sensitizing used in this study was 3.5\u0026nbsp;mm/s, which was sufficient to partially attenuate the CSF-signal in PVS, indicating that a part of CSF in PVS moves faster than 3.5 mm/s (Suppl. Fig. 1). Since the current sequence does not allow for measuring phase-accrual generated by the motion-sensitizing gradients, it is not possible to determine whether CSF exhibits net flow in the PVS or if the CSF-mobility in PVS would be better described as a mixing phenomenon: the direction (towards one end of the axis of movement or the other, e.g. right-to-left or left-to-right) cannot be determined with the current sequence. Still, a potential propagation pattern of CSF-mobility across cardiac phases can be observed (Fig. 3, Fig. 4, Suppl. video 2, Suppl. video 3), suggesting that CSF-mobility varies in distinct spatially coherent waves driven by the cardiac and respiratory cycle. This could further enhance the efflux of waste products from the neuropil to the SAS.\u003c/p\u003e\n\u003cp\u003eA\u0026nbsp;further limitation of the current approach is the long scan duration that leads to\u0026nbsp;a higher\u0026nbsp;sensitivity to motion (two datasets had to be excluded due to motion artefacts) and to a low intrinsic temporal resolution, which limits the study of pulsatile dynamics that cannot be predicted in advance or using an external trigger (e.g. vasomotion, as opposed to cardiac or respiratory dynamics). Regarding the scan duration, the current k-space sampling rendered sufficient SNR, allowing to further split k-space to study the effect of cardio-respiratory pulsations. This suggests that for baseline CSF-mobility measurements (i.e. without retrospective binning), further acceleration can be achieved. Shorter acquisition times resulting from such higher acceleration combined with the use of prospective motion correction, e.g. by navigators\u003csup\u003e37,38\u003c/sup\u003e, would also help reduce sensitivity to motion. However, even with the current scan-times it was possible to identify CSF-mobility changes in patients with\u0026nbsp;a\u0026nbsp;neurodegenerative disease.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe truly non-invasive nature of the developed technique (as opposed to techniques using contrast agent injections) allows the inclusion of this methodology into longitudinal, patient, and population studies. It also makes the sequence repeatable (i.e. applicable several times in a row), which would be essential when applied to sleep research or to monitor disease progression for example in neurodegenerative diseases. The information obtained from such studies could lead to important insights into the ways human brain pathologies are affected by impaired brain clearance and lead to new ways of improving brain health.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eCSF-STREAM enables detailed measurement of CSF-mobility in humans, from large CSF-filled spaces down to PVS surrounding penetrating arteries in a fully non-invasive manner. Cardiac and respiratory fluctuations induced comparable oscillations in PVS, whereas in larger CSF-spaces at the base of the brain, the cardiac cycle was found to be the main driving force of CSF-mobility. Moreover, regional alterations in CSF-mobility were found in patients with a presumed brain clearance disorder. Finally, CSF-mobility in the visual cortex could be enhanced through entraining vasomotion at 0.1 Hz.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003eSubjects\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eHealthy, younger cohort\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eA total of 24 healthy individuals (age: 33\u0026plusmn;13 years, 20 females, 4 males) were scanned: 14 individuals were enrolled in the first study to evaluate CSF-mobility fluctuations across cardiac/respiration/random phases. One individual was excluded because of motion artefacts and two because of insufficient quality of the cardiac signal.\u0026nbsp;Motion\u0026nbsp;was identified\u0026nbsp;in the reconstructed images as blurring\u0026nbsp;of the images and\u0026nbsp;duplication of\u0026nbsp;brain\u0026nbsp;structures.\u0026nbsp;Ten participated in the second study to investigate the effect of a visual stimulation on CSF-mobility, of which one\u0026nbsp;was\u0026nbsp;excluded due to motion artefacts. Six individuals participated in both studies. All were screened for MRI contra-indications and provided written informed consent. All experiments were performed in accordance with the Leiden University Medical Center Institutional Review Board (Leiden, The Netherlands) under authorization number P07.096.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCAA cohort\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eCAA patients (n=8) were recruited through the neurovascular outpatient clinic at University Hospital Bonn. The diagnosis of probable CAA was independently confirmed by a board-certified neuroradiologist according to the Boston Criteria version 2.0\u003csup\u003e29\u003c/sup\u003e. Age- and sex-matched healthy participants (n=8) were recruited through the DANCER cohort, i.e. a neurologically unaffected control cohort of the German Center for Neurodegenerative Diseases (DZNE). The study was approved by the Ethics Committee of University Hospital Bonn. Written informed consent was obtained for each participant.\u003c/p\u003e\n\u003cp\u003eMedical history was obtained from each participant to record relevant previous diseases, cardiovascular risk factors, degree of disability, lifestyle factors and physical activity. Degree of disability was assessed using the modified Rankin Scale. Physical activity was assessed using the physical activity scale for the elderly (PASE). Global cognitive status was assessed using the Montreal cognitive assessment test. Trail Making Tests (TMT) A and B were used to assess executive function and processing speed. Symptoms of depression were assessed using the Center for Epidemiologic Depression Scale (CES-D), revised Becker Depression Inventory (BDI-II) and Geriatric Depression Scale (GDS).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMRI scans acquisition\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eHealthy, younger cohort\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAll scans were acquired using a 7 Tesla MRI scanner (Achieva, Philips, Best, The Netherlands) equipped with a quadrature birdcage head coil and a 32-channel receive coil array (Nova Medical, Wilmington, MA, USA).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAnatomical 3D-T\u003csub\u003e1\u003c/sub\u003e scan\u003c/em\u003e: 3D T\u003csub\u003e1\u003c/sub\u003e-weighted images were acquired using the following parameters: field-of-view = 246\u0026times;246\u0026times;225 mm\u003csup\u003e3\u003c/sup\u003e, flip angle = 7\u003csup\u003e◦\u003c/sup\u003e,\u0026nbsp;echo time (TE) = 1.9\u0026nbsp;ms, repetition time (TR) = 4.2\u0026nbsp;s, spatial resolution = 0.9\u0026nbsp;mm isotropic, and acquisition time = 142\u0026nbsp;s.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCSF-STREAM\u003c/em\u003e: High-resolution, whole-brain (0.45 mm isotropic voxel-size, field-of-view = 250\u0026times;250\u0026times;190 mm), 3D images were acquired with a\u0026nbsp;TSE\u0026nbsp;sequence (TE = 495 ms, TR = 3.4 s, TSE-factor 146, excitation \u0026amp; refocusing FA = 90\u003csup\u003e◦\u003c/sup\u003e). This long echo-time readout was combined with a T\u003csub\u003e2\u003c/sub\u003e-preparation module (duration = 37 ms, 2 refocusing pulses) in order to allow the insertion of motion sensitizing gradients and to further isolate the CSF-signal.\u003c/p\u003e\n\u003cp\u003eTo accelerate the acquisition, k-space undersampling in \u003cem\u003eky\u003c/em\u003e and \u003cem\u003ekz\u003c/em\u003e phase encoding directions compatible with compressed sensing reconstruction was performed using the Amsterdam UMC PROUD patch\u003csup\u003e39\u003c/sup\u003e, based on a pseudo-radial variable density (density decay = 0.5) sampling pattern with a fully sampled 29\u0026times;29 auto-calibration area in the center of k-space. This way, an acceleration factor of 17 was achieved, which allowed to obtain high spatial resolution, whole-brain, static CSF-images in 5 min 30 s. Subsequently, motion-sensitizing gradients were included in the T\u003csub\u003e2\u003c/sub\u003e-preparation in order to encode CSF-mobility. Seven sets of volumes (sub-scans) were acquired: one without motion-sensitizing gradients and six with gradients applied in different, orthogonal directions. In practice, motion-sensitized gradients of 5\u0026nbsp;mm/s were played out on two axes simultaneously, resulting in a diagonal direction with a motion-encoding of 5/\u0026radic;2 = 3.5\u0026nbsp;mm/s. The acquisition time per sub-scan was 5 min 30 s, yielding a total scan time of 38 min 30 s.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eVisual stimulation scout fMRI scan\u003c/em\u003e: A visual scout fMRI blood oxygen level dependent (BOLD) scan was acquired in order to locate the visual cortex. The acquisition parameters were as follows: field-of-view = 222\u0026times;190 mm\u003csup\u003e2\u003c/sup\u003e, flip angle = 70\u003csup\u003e◦\u003c/sup\u003e,\u0026nbsp;TE = 22\u0026nbsp;ms, TR = 2\u0026nbsp;s, 35 slices,\u0026nbsp;voxel size:\u0026nbsp;1.97\u0026times;1.74\u0026nbsp;mm\u003csup\u003e2\u003c/sup\u003e in-plane, 2 mm slice thickness, EPI-factor = 43, 60 repetitions (timepoints), acquisition time = 128 s. The visual stimulus consisted of 3 blocks of an 8 Hz flashing radial black-and-white checkerboard pattern for 20 seconds alternated with 20 seconds of a fixed grey screen as rest condition.\u003c/p\u003e\n\u003cp\u003eUsing the above-mentioned scans, two studies were performed\u0026nbsp;in the healthy, younger cohort:\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eStudy 1: CSF-mobility across cardiac, respiration and random phases \u0026ndash;\u003c/em\u003e This protocol consisted of an anatomical 3D-T\u003csub\u003e1\u003c/sub\u003e scan, a CSF-STREAM scan (one scan = one non-motion-sensitized sub-scan + 6 motion-sensitized sub-scans) and a visual stimulation scout scan (except for the first four subjects). During the acquisition, the heart rate was continuously recorded using a peripheral pulse unit and the respiration rate using a belt wrapped around the subject\u0026rsquo;s chest; both physiological monitoring devices were the standard equipment as supplied by the vendor (Philips, Best, The Netherlands).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eStudy 2: entrained vasomotion\u003c/em\u003e \u003cem\u003eusing a 0.1 Hz visual stimulation \u0026ndash;\u0026nbsp;\u003c/em\u003eThis protocol consisted of an anatomical 3D-T\u003csub\u003e1\u003c/sub\u003e scan, a CSF-STREAM scan and a visual stimulation scout scan. During each of the seven CSF-STREAM sub-scans, a visual stimulus was shown during the first half of the sub-scan, and a grey screen during the second half of the sub-scan. The stimulation paradigm consisted of an 8 Hz flashing radial black and white checkerboard pattern for 5 s alternated with 5 s of a fixed grey screen, altogether leading to a 0.1 Hz stimulation frequency. In order to sample k-space homogeneously in both sub-scan halves, k-space sampling was readjusted by first acquiring the odd TSE-shots of the original k-space sampling in the first half and subsequently the even TSE-shots during the second half. \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCAA cohort\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eMRI data were acquired at a 7 Tesla MRI system (Siemens\u0026nbsp;Healthineers, Erlangen, Germany) using\u0026nbsp;a head array coil with 32 receive and 8 transmit channels (Nova Medical, Wilmington, MA, USA). Scans were performed\u0026nbsp;in the morning between 09:30 am and 11:30 am.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe scan protocol included a\u0026nbsp;T\u003csub\u003e1\u003c/sub\u003e-weighted multi-echo MPRAGE (0.80 mm isotropic, TR\u0026nbsp;= 2800 ms, TI\u0026nbsp;= 1100 ms, scan time = 4 min) and a CSF-STREAM scan (0.50 mm isotropic, TE = 515\u0026nbsp;ms, TR\u0026nbsp;= 3400 ms, TSE-factor 146, refocusing FA\u0026nbsp;= 70\u003csup\u003e◦\u003c/sup\u003e, 12x Poisson-disk undersampling scheme, acquisition time 4:15 min per sub-scan, motion-sensitizing gradients of 4\u0026nbsp;mm/s). In practice, similar to the study in healthy, young volunteers, motion-sensitized gradients of 5.6\u0026nbsp;mm/s were played out on two axes simultaneously, resulting in a diagonal direction with a motion-encoding of 4\u0026nbsp;mm/s. Additionally, susceptibility-weighted images (0.50 mm isotropic) were acquired to quantify cerebral microbleeds\u0026nbsp;and to identify superficial siderosis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCSF-STREAM image reconstruction\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eHealthy, younger cohort\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAll CSF-STREAM reconstructions\u0026nbsp;of this cohort\u0026nbsp;were performed offline in MATLAB 2018b (The Mathworks, USA), using an in-house built reconstruction pipeline developed within ReconFrame (GyroTools, Z\u0026uuml;rich, Switzerland) in combination with the open-source Berkeley Advanced Reconstruction Toolbox (BART)\u003csup\u003e40\u003c/sup\u003e version 0.4.03.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eStudy 1: retrospective cardiac, respiration and random binning -\u0026nbsp;\u003c/em\u003eAfter acquisition, each of the seven CSF-STREAM sub-scans was reconstructed three times: the k-space profiles were retrospectively binned in 6 phases, using retrospective binning to either (1) the cardiac cycle, (2) the respiratory cycle or (3) to random phases (negative control). The R-R cardiac peaks as well as the respiration peaks were detected automatically from the recorded cardiac and respiratory traces using a MATLAB script (using the \u003cem\u003efindpeaks\u003c/em\u003e function) and the results were checked manually for each scan and corrected if wrongly detected. Random phases were generated using the \u003cem\u003erandi\u003c/em\u003e MATLAB function. The retrospective binning in 6 phases was performed in two steps in order to preserve image quality: k-space was binned twice in 3 phases, with a 1/6\u003csup\u003eth\u003c/sup\u003e phase shift between the two steps, and subsequently combined in one dataset (Extended data Fig. 3). The T\u003csub\u003e2\u003c/sub\u003e-preparation preceding each TSE-shot was taken as reference and the signal in the subsequent TSE-shot was considered to be dependent on this specific cardiac/respiration/random phase\u003csup\u003e41,42\u003c/sup\u003e. Each reconstruction step was performed using BART\u0026rsquo;s \u003cem\u003epics\u003c/em\u003e command with total variation in the temporal domain with a regularization factor of 0.005 and input coil sensitivities estimated from the k-space center using BART\u0026rsquo;s \u003cem\u003ecaldir\u003c/em\u003e command. Altogether, this generated 6 tensors per driving force per subject (i.e. one tensor per cardiac/respiration/random phase).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eStudy 2: \u0026nbsp; 0.1 Hz visual stimulation\u0026nbsp;\u003c/em\u003e\u0026ndash; Thetwo half k-spaces (0.1 Hz stimulation versus rest) of each of the seven CSF-STREAM sub-scans were reconstructed using BART\u0026rsquo;s \u003cem\u003epics\u003c/em\u003e command with total variation in the temporal domain with a regularization factor of 0.005 and input coil sensitivities estimated from the k-space center using BART\u0026rsquo;s \u003cem\u003ecaldir\u003c/em\u003e command. Per subject, this resulted in a set of two tensors: one with 0.1 Hz stimulation and one at rest.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCAA cohort\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eImage reconstruction of CSF-STREAM sub-scans was achieved using the \u003cem\u003epics\u003c/em\u003e command from the BART Toolbox with l1-regularization (regularization\u0026nbsp;factor\u0026nbsp;0.002) and 30 iterations. Input coil sensitivities were estimated from a fully-sampled gradient-echo pre-scan using BART\u0026rsquo;s \u003cem\u003eecalib\u003c/em\u003e command.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePost-processing \u0026ndash; healthy, younger cohort\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCSF-mobility, FA and principal orientation of CSF-mobility\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eIntra-subject images were registered using Elastix version 4.9.0\u003csup\u003e43\u003c/sup\u003e. CSF-mobility, its principal orientation, and FA were modelled using a MATLAB script\u003csup\u003e44\u003c/sup\u003e by computing the mean eigenvalue of a rank-two positive definite tensor, analogous to\u0026nbsp;DTI. CSF-mobility is a measure of the amount of movement that CSF undergoes within a given time in a voxel as a function of the intra-voxel dephasing of signal due to the application of bipolar gradients. It is measured in units of mm\u003csup\u003e2\u003c/sup\u003e/s and is calculated in a similar way as an apparent diffusion coefficient.\u003c/p\u003e\n\u003cp\u003eFor the first study that investigated the effect of the cardiac/respiration/random pulsations, six CSF-mobility and FA maps were created (one for each phase) per driving force per subject. CSF-mobility change and FA-change maps were obtained for each cardiac, respiration and random phase after normalizing the maps voxel-wise to the mean value over phases.\u003c/p\u003e\n\u003cp\u003eCSF-mobility and FA changes across driving forces were investigated in six ROIs. As can be seen in Fig. 3 and Fig. 4, the fluctuations across phases vary spatially, i.e. the phase of maximum signal is location-dependent. To avoid phase cancellation within an ROI and hence potential smoothing of the effect of a driving force, the time-profiles were realigned voxel-wise before calculating the average CSF-mobility and FA changes in an ROI. That way, phase 1 always contained the maximum signal change and was discarded from the plots in Fig. 5 and Extended data Fig. 4.\u003c/p\u003e\n\u003cp\u003eTo quantitatively compare how CSF-mobility varies across the proposed driving forces, the\u0026nbsp;original\u0026nbsp;CSF-mobility change\u0026nbsp;over 6 phases (i.e. without the realigment mentioned in the previous paragraph)\u0026nbsp;was fitted voxel-wise to a sine function. The fit quality, the maximum amplitude as well as the phase of the maximum amplitude were the fitted parameters. The amplitude of voxels with a low fit quality (R\u003csup\u003e2\u003c/sup\u003e\u0026lt;0.5)\u0026nbsp;was set to 0\u0026nbsp;to better represent the absence of coherent change across phases. For example, in the right panel of Extended Fig. 6, the fitted curve should ideally be flat (i.e. amplitude of 0% instead of 12%) because the signal pattern is more noisy (\u0026ldquo;zig-zag\u0026rdquo;) than coherent across phases.\u003c/p\u003e\n\u003cp\u003eFor the second study investigating the effect of entrained vasomotion on CSF-mobility, two CSF-mobility maps were created: one acquired during the 0.1 Hz visual stimulation and another at rest. The relative difference between the average CSF-mobility during the 0.1 Hz\u0026nbsp;stimulation\u0026nbsp;and rest (no stimulation) was computed to investigate the effect of entrained vasomotion. In the subjects who participated in both studies, the effect induced by entrained vasomotion was compared to that from cardiac, respiratory, and random cycles (Fig. 7D) by computing the difference between the maximum and minimum CSF-mobility change over cardiac/respiration/random phases.\u003c/p\u003e\n\u003cp\u003eAll results were visualized using MATLAB and Paraview\u003csup\u003e45\u003c/sup\u003e. When plotting CSF-mobility/FA volume renderings or CSF-mobility change maps, voxels where the CSF-signal was lower than 150 were masked to exclude noise. The CSF-mobility orientation plots were visualized in smaller volumes of interest. Before computing the CSF-mobility orientation as described above, the resolution of the small volume of interest of the seven CSF-STREAM sub-scans was doubled using the MATLAB function \u003cem\u003eimresize3\u003c/em\u003e, in order to better show the orientation in the selected regions.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003efMRI processing\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003ez-score map generation\u003c/em\u003e: fMRI data processing was carried out using FEAT (FMRI Expert Analysis Tool) Version 6.00, part of FSL. The following pre-statistics processing were applied: motion correction using MCFLIRT\u003csup\u003e46\u003c/sup\u003e; slice-timing correction using Fourier-space time-series phase-shifting; non-brain removal using the Brain Extraction Tool (BET)\u003csup\u003e47\u003c/sup\u003e; spatial smoothing using a Gaussian kernel of FWHM 3 mm; grand-mean intensity normalization of the entire 4D dataset by a single multiplicative factor; high-pass temporal filtering (Gaussian-weighted least-squares straight line fitting, with sigma = 50s). To investigate the possible presence of unexpected artefacts or activation, ICA-based exploratory data analysis was carried out using MELODIC\u003csup\u003e48\u003c/sup\u003e. The statistical analysis of the time-series was carried out using FILM with local autocorrelation correction\u003csup\u003e49\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003ez-score template (study 1):\u003c/em\u003e As the first four subjects of study 1 did not have a visual stimulation scout scan, a mean z-score template was generated to create a visual cortex mask for study 1, using the scout scan of a subset of ten subjects (from study 1 and 2; only one scout scan per subject was included, i.e. scans from the six subjects included in both studies were only included once). To create this z-score template, the ten BOLD scans were registered to the 3D-T\u003csub\u003e1\u003c/sub\u003e scans using a boundary based registration\u003csup\u003e50\u003c/sup\u003e. This transformation was applied to the z-score maps. Then, 3D-T\u003csub\u003e1\u003c/sub\u003e scans \u0026nbsp; were registered to MNI-space using FSL\u0026rsquo;s FLIRT\u003csup\u003e46,51\u003c/sup\u003e and FNIRT\u003csup\u003e52\u003c/sup\u003e. The resulting warpfield was applied to the registered z-score maps. The template was then generated using a one-sample group mean generalized linear model within FreeSurfer\u003csup\u003e53\u003c/sup\u003e and transformed into the CSF-mobility space of each individual subject of study 1.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eIndividual z-score map (study 2):\u003c/em\u003e For study 2, a visual stimulation scout scan was available for each subject. Therefore, for this study, individual z-score maps were used to best detect the visual cortex in each subject. The person-specific z-score maps were thus registered to each individual CSF-mobility space in the following way. First, the BOLD scans were registered to the 3D-T\u003csub\u003e1\u003c/sub\u003e scans using a boundary based registration\u003csup\u003e50\u003c/sup\u003e. Subsequently, the 3D-T\u003csub\u003e1\u003c/sub\u003e scans were skull stripped and segmented in brain tissue types\u003csup\u003e54\u003c/sup\u003e. The obtained CSF probability map was registered to CSF-mobility space with an Euler registration using Elastix\u003csup\u003e43\u003c/sup\u003e. The resulting transformations were then applied to the z-score maps.\u003c/p\u003e\n\u003cp\u003eThe obtained z-score maps were used to detect the visual cortex.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe BOLD signal amplitude was calculated from the registered, motion corrected, slice-time corrected BOLD images. The 3 stimulation patterns (3\u0026times;20 dynamics) were first averaged together, then the relative difference between baseline signal (averaged over dynamics 5-10) and maximum signal (averaged over dynamics 15-20) was computed.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eROI definition\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eROIs were manually delineated using anatomical landmarks on the non-motion-sensitized CSF-scan (cf. Fig. 2) using ITK-Snap\u003csup\u003e55\u003c/sup\u003e, as follows:\u0026nbsp;\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eThe 4\u003csup\u003eth\u003c/sup\u003e ventricle ROI was drawn over 13 transversal slices\u003c/li\u003e\n \u003cli\u003eThe ROI delimiting the SAS around the MCA was drawn over seven transversal slices on the left MCA branch\u003c/li\u003e\n \u003cli\u003eThe motor cortex SAS sulci ROI was drawn over ten transversal slices\u003c/li\u003e\n \u003cli\u003ePVS in the basal ganglia were identified over sagittal slices as CSF-filled spaces around the lenticulostriate arteries\u003c/li\u003e\n \u003cli\u003eThe ROI of PVS surrounding penetrating arteries in the white matter was drawn over 25 transversal slices in the centrum semiovale, starting from the slice directly above the lateral ventricles\u003c/li\u003e\n \u003cli\u003eThe blood ROI was drawn inside one MCA branch over two slices\u003c/li\u003e\n \u003cli\u003eThe noise ROI was drawn outside the brain, on the two central sagittal slices and on the corners of the same slice where the SAS MCA ROI was drawn\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eFor study 1, the visual cortex ROI was defined based on the template z-score output (four datasets of study 1 did not contain a visual stimulation scout scan). A threshold of z-score\u0026gt;3.5 was used for all subjects of this study. Only the largest cluster of contiguous voxels was included in the mask. As the resolution of the fMRI scan used to create the z-score map was much lower than that of the CSF-STREAM, the obtained area not only contained the visual cortex but also the CSF in its vicinity.\u003c/p\u003e\n\u003cp\u003eTo extract the final areas of interest, the manually delineated ROIs and the visual cortex ROI were multiplied with a CSF-mask, thus including only voxels containing CSF and not noise. This CSF-mask was created by first thresholding the non-motion-sensitized CSF-scan using a threshold of 150 (a.u.). After visual inspection, this threshold could be adapted individually to assure proper selection of PVS. For study 1, if a voxel had a CSF-mobility change higher than 50% in one or more of the cardiac/respiration/random datasets, it was excluded from the mask. Voxels that had no included neighboring voxels (\u0026ldquo;lonely\u0026rdquo; voxels) were also excluded from the mask. For the two PVS ROIs, an additional Frangi filter\u003csup\u003e56\u003c/sup\u003e (0.6\u0026lt;\u0026sigma;\u0026lt;1 with a step of 0.2, Frangi vesselness constant = 0.5) was applied in order to ensure the exclusive inclusion of vessel-like structures and exclusion of noise.\u003c/p\u003e\n\u003cp\u003eFor study 2, a visual stimulation scout scan was available for all subjects, and we therefore used individual z-score maps to create the visual cortex ROIs. A threshold of z-score\u0026gt;7 was used to define the visual area and a threshold of z-score\u0026lt;1 was used to define a control region (rest of the brain). For the visual cortex ROI, only the largest cluster of contiguous voxels was included in the ROI. Next, to ensure only voxels containing CSF were included and not noise, a mask based on the non-motion-sensitized CSF-scan was created using a threshold of 150\u0026nbsp;(a.u.). Voxels for which the CSF-mobility change between the 2 conditions (stimulation ON and OFF) was higher than 50% were considered as noise and excluded from the ROI. As the resolution of the fMRI scan used to create the z-score map was much lower than that of the CSF-STREAM, the obtained area not only contained the visual cortex but also the CSF around the visual cortex\u0026nbsp;(SAS and PVS).\u003c/p\u003e\n\u003cp\u003eThe effect induced by the cardiac and respiratory cycles was compared to the changes induced by entrained vasomotion in the subjects who participated in both studies. To that end, the personalized visual area mask created for study 2 was registered to the CSF-space of study 1 using Elastix.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePost-processing - CAA cohort\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCSF-mobility and FA\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe CSF-STREAM sub-scans were first interpolated from a 0.50 mm to a 0.17 mm isotropic resolution and co-registered using Elastix. Subsequently, the mean eigenvalue of a rank-two positive definite tensor was computed (DTI post-processing) using Python 3.10to assess CSF-mobility\u0026nbsp;and\u0026nbsp;FA. To minimize the effects of background noise, a cut-off value of 0.15 mm\u0026sup2;/s was used to exclude all voxels with unphysiologically high CSF-mobility values.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eROI definition\u003c/em\u003e\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eSAS-MCA segmentations: the M1 segment of the middle cerebral artery was first segmented semi-manually using anatomical landmarks on the non-motion-sensitized CSF-scan using MITK version v2022.10. The MCA-segmentation was then inflated using the \u003cem\u003eimdilate\u003c/em\u003e Matlab function to create a CSF-mask containing the SAS around the MCA; inflation was done from 0.17\u0026nbsp;mm up to 3.00 mm with a step-size of 0.17\u0026nbsp;mm. To ensure only CSF-signal was selected in the mask, voxels with low CSF-signal in the non-motion sensitized scan were excluded.\u003c/li\u003e\n \u003cli\u003ePVS segmentations: Parcellated atlases from the\u0026nbsp;T\u003csub\u003e1\u003c/sub\u003e-scan\u0026nbsp;were generated using Freesurfer (Version 6.0). White matter segmentations, derived from the Freesurfer parcellation, were manually corrected if necessary. To capture comparable ROIs of the CSO in all participants, the eyes and the optic chiasm served as anatomical landmarks for reference plane definition. The dimensions of the CSO segmentation encompassed the entire white matter above the lateral ventricles. PVS within the defined CSO segmentation were semiautomatically segmented using a Meijering filter-based approach\u003csup\u003e57\u003c/sup\u003e with a global threshold on the interpolated non-motion sensitized scan of CSF-STREAM (0.17 mm isotropic).\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u003cem\u003eAssessment of microbleeds\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eCerebral microbleeds were quantified\u0026nbsp;and superficial siderosis was identified\u0026nbsp;by a board-certified neuroradiologist with 8 years of experience\u0026nbsp;according to the STRIVE-2 rating scale.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical analysis\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWithin each ROI, we evaluated whether the fit quality as well as the amplitude of CSF-mobility change were significantly different between driving forces. A post-hoc Bonferroni-corrected pairwise comparison between driving force effects was performed using a Wilcoxon signed rank test when Friedman\u0026rsquo;s test was significant. To evaluate the effect of the visual stimulation, a Wilcoxon signed rank test was performed on the average CSF-mobility values with and without stimulation from each subject. These above-mentioned statistical analyses were performed in MATLAB.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo evaluate the evidence for a correlation between the BOLD amplitude and CSF-mobility change with a visual stimulation, a Bayesian correlation analysis was performed using JASP software\u003csup\u003e58\u003c/sup\u003e (JASP Team (2022). JASP (Version 0.16.4)). A stretched beta prior with a width of 1.0 was used, and the Bayes Factor was tested to be stable over a range of prior settings.\u003c/p\u003e\n\u003cp\u003eTo evaluate differences in CSF-mobility, FA, ROI volume and patient information between CAA and healthy controls, Mann-Whitney U-tests were performed.\u003c/p\u003e\n\u003cp\u003eUnless mentioned otherwise, shaded error areas and error bars represent confidence intervals of SD\u0026times;1.96/\u0026radic;n, n being the number of included subjects.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eBoxplots were plotted using the MATLAB \u003cem\u003eboxplot\u003c/em\u003e function: on each boxplot, the central line indicates the median, and the bottom and top edges of the box indicate the 25\u003csup\u003eth\u003c/sup\u003e and 75\u003csup\u003eth\u003c/sup\u003e percentiles, respectively. The whiskers extend to the most extreme data points not considered outliers, and the outliers are plotted individually using the \u0026apos;+\u0026apos; marker symbol.\u003c/p\u003e\n\u003cp\u003ePreliminary parts of this work were presented at conferences\u003csup\u003e41,42,59,60\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements:\u0026nbsp;\u003c/strong\u003eThe authors would like to thank Lukas Gottwald and Aart Nederveen for providing the Amsterdam UMC PROUD patch, Thomas Roos and Kai L\u0026oslash;nning for help with the reconstructions, Geir Ringstad for fruitful discussions and R\u0026uuml;diger Stirnberg for his support and input with regards to the 7 Tesla MRI protocol. This work is part of the research program Innovational Research Incentives Scheme Vici with project number 016.160.351, which is financed by the Netherlands Organization for Scientific Research (NWO). It was furthermore supported by the Women in MR award from the ISMRM-Benelux, Alzheimer Nederland (Young Outstanding Researcher Award WE.25-2020-05 to Susanne van Veluw and travel grant to Lydiane Hirschler),\u0026nbsp;the Joint Program for Neurodegenerative Diseases (JPND) on Human Brain Clearance Imaging (HBCI), the Federal Ministry of Education and Research (BMBF) in Germany (funding code 01KX2130), as well as the\u0026nbsp;Leducq Foundation (Transatlantic Network of Excellence 23CVD03). Katerina Deike was funded by the Medical Faculty of University Bonn (2022-FKS-02, 2024-FKS-02 and Femhabil 03-2022). Gabor Petzold received funding from\u0026nbsp;the DZNE and Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany\u0026rsquo;s Excellence Strategy \u0026ndash; EXC2151 \u0026ndash; 390873048. Daniel Paech was funded by a grant (2022-EKES.33) of the Else Kr\u0026ouml;ner-Fresenius-Stiftung (EKFS), Philipp Vollmuth is supported through an Else Kr\u0026ouml;ner Clinician Scientist Endowed Professorship by the Else Kr\u0026ouml;ner Fresenius Foundation (reference number: 2022_EKCS.17) and Alexander Effland was funded by the German Research Foundation under Germany\u0026rsquo;s Excellence Strategy EXC-2047/1\u0026ndash;390685813 and EXC2151\u0026ndash;390873048.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;of Interest:\u003c/strong\u003e The LUMC receives research support from Philips,\u0026nbsp;M.W.A. Caan is a shareholder of Nicolab International Ltd,\u0026nbsp;Katerina Deike\u0026nbsp;and Daniel Paech are\u0026nbsp;co-founders\u0026nbsp;and shareholders\u0026nbsp;of relios.vision GmbH. Daniel Paech is on the Guerbet advisory board.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability:\u003c/strong\u003e Data will be made available upon reasonable request and with a data exchange agreement.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode availability:\u003c/strong\u003e Codes are available upon reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eDolgin, E. Brain\u0026rsquo;s drain. \u003cem\u003eNat. Biotechnol.\u003c/em\u003e (2020) doi:10.1038/s41587-020-0443-1.\u003c/li\u003e\n\u003cli\u003eRasmussen, M. K., Mestre, H. \u0026amp; Nedergaard, M. Fluid Transport in the Brain. \u003cem\u003ePhysiol. Rev.\u003c/em\u003e (2021) doi:10.1152/physrev.00031.2020.\u003c/li\u003e\n\u003cli\u003eIliff, J. J. \u003cem\u003eet al.\u003c/em\u003e A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid \u0026beta;. \u003cem\u003eSci. Transl. Med.\u003c/em\u003e (2012) doi:10.1126/scitranslmed.3003748.\u003c/li\u003e\n\u003cli\u003eMestre, H. \u003cem\u003eet al.\u003c/em\u003e Flow of cerebrospinal fluid is driven by arterial pulsations and is reduced in hypertension. \u003cem\u003eNat. Commun.\u003c/em\u003e (2018) doi:10.1038/s41467-018-07318-3.\u003c/li\u003e\n\u003cli\u003evan Veluw, S. J. \u003cem\u003eet al.\u003c/em\u003e Vasomotion as a Driving Force for Paravascular Clearance in the Awake Mouse Brain. \u003cem\u003eNeuron\u003c/em\u003e (2020) doi:10.1016/j.neuron.2019.10.033.\u003c/li\u003e\n\u003cli\u003eRingstad, G., Vatnehol, S. A. S. \u0026amp; Eide, P. K. Glymphatic MRI in idiopathic normal pressure hydrocephalus. \u003cem\u003eBrain\u003c/em\u003e (2017) doi:10.1093/brain/awx191.\u003c/li\u003e\n\u003cli\u003eRingstad, G. \u003cem\u003eet al.\u003c/em\u003e Brain-wide glymphatic enhancement and clearance in humans assessed with MRI. \u003cem\u003eJCI insight\u003c/em\u003e (2018) doi:10.1172/jci.insight.121537.\u003c/li\u003e\n\u003cli\u003eWardlaw, J. M. \u003cem\u003eet al.\u003c/em\u003e Perivascular spaces in the brain: anatomy, physiology and pathology. \u003cem\u003eNat. Rev. Neurol.\u003c/em\u003e (2020) doi:10.1038/s41582-020-0312-z.\u003c/li\u003e\n\u003cli\u003eBakker, E. N. T. P., Naessens, D. M. P. \u0026amp; VanBavel, E. Paravascular spaces: entry to or exit from the brain? \u003cem\u003eExp. Physiol.\u003c/em\u003e (2019) doi:10.1113/EP087424.\u003c/li\u003e\n\u003cli\u003eMestre, H., Mori, Y. \u0026amp; Nedergaard, M. The Brain\u0026rsquo;s Glymphatic System: Current Controversies. \u003cem\u003eTrends Neurosci.\u003c/em\u003e \u003cstrong\u003exx\u003c/strong\u003e, 1\u0026ndash;9 (2020).\u003c/li\u003e\n\u003cli\u003eAldea, R., Weller, R. O., Wilcock, D. M., Carare, R. O. \u0026amp; Richardson, G. Cerebrovascular smooth muscle cells as the drivers of intramural periarterial drainage of the brain. \u003cem\u003eFront. Aging Neurosci.\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, (2019).\u003c/li\u003e\n\u003cli\u003eZhao, L., Tannenbaum, A., Bakker, E. N. T. P. \u0026amp; Benveniste, H. Physiology of Glymphatic Solute Transport and Waste Clearance from the Brain. \u003cem\u003ePhysiology (Bethesda).\u003c/em\u003e \u003cstrong\u003e37\u003c/strong\u003e, 0 (2022).\u003c/li\u003e\n\u003cli\u003eVinje, V. \u003cem\u003eet al.\u003c/em\u003e Respiratory influence on cerebrospinal fluid flow \u0026ndash; a computational study based on long-term intracranial pressure measurements. \u003cem\u003eSci. Rep.\u003c/em\u003e (2019) doi:10.1038/s41598-019-46055-5.\u003c/li\u003e\n\u003cli\u003eRasmussen, M. K., Mestre, H. \u0026amp; Nedergaard, M. The glymphatic pathway in neurological disorders. \u003cem\u003eThe Lancet Neurology\u003c/em\u003e (2018) doi:10.1016/S1474-4422(18)30318-1.\u003c/li\u003e\n\u003cli\u003eGreenberg, S. M. \u003cem\u003eet al.\u003c/em\u003e Cerebral amyloid angiopathy and Alzheimer disease \u0026mdash; one peptide, two pathways. \u003cem\u003eNature Reviews Neurology\u003c/em\u003e (2020) doi:10.1038/s41582-019-0281-2.\u003c/li\u003e\n\u003cli\u003eChristensen, J., Wright, D. K., Yamakawa, G. R., Shultz, S. R. \u0026amp; Mychasiuk, R. Repetitive Mild Traumatic Brain Injury Alters Glymphatic Clearance Rates in Limbic Structures of Adolescent Female Rats. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, (2020).\u003c/li\u003e\n\u003cli\u003eMestre, H. \u003cem\u003eet al.\u003c/em\u003e Cerebrospinal fluid influx drives acute ischemic tissue swelling. \u003cem\u003eScience (80-. ).\u003c/em\u003e (2020) doi:10.1126/science.aax7171.\u003c/li\u003e\n\u003cli\u003evan Veluw, S. J. \u003cem\u003eet al.\u003c/em\u003e Is CAA a perivascular brain clearance disease? A discussion of the evidence to date and outlook for future studies. \u003cem\u003eCell. Mol. Life Sci.\u003c/em\u003e \u003cstrong\u003e81\u003c/strong\u003e, (2024).\u003c/li\u003e\n\u003cli\u003eChen, X. \u003cem\u003eet al.\u003c/em\u003e Cerebral amyloid angiopathy is associated with glymphatic transport reduction and time-delayed solute drainage along the neck arteries. \u003cem\u003eNat. Aging\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003cli\u003eAlbargothy, N. J. \u003cem\u003eet al.\u003c/em\u003e Convective influx/glymphatic system: tracers injected into the CSF enter and leave the brain along separate periarterial basement membrane pathways. \u003cem\u003eActa Neuropathol.\u003c/em\u003e (2018) doi:10.1007/s00401-018-1862-7.\u003c/li\u003e\n\u003cli\u003eJanssen, P. M. L., Biesiadecki, B. J., Ziolo, M. T. \u0026amp; Davis, J. P. The need for speed: Mice, men, and myocardial kinetic reserve. \u003cem\u003eCirc. Res.\u003c/em\u003e \u003cstrong\u003e119\u003c/strong\u003e, (2016).\u003c/li\u003e\n\u003cli\u003eBito, Y., Harada, K., Ochi, H. \u0026amp; Kudo, K. Low b-value diffusion tensor imaging for measuring pseudorandom flow of cerebrospinal fluid. \u003cem\u003eMagn. Reson. Med.\u003c/em\u003e \u003cstrong\u003e86\u003c/strong\u003e, (2021).\u003c/li\u003e\n\u003cli\u003eWen, Q. \u003cem\u003eet al.\u003c/em\u003e Assessing pulsatile waveforms of paravascular cerebrospinal fluid dynamics using dynamic diffusion‐weighted imaging (dDWI). \u003cem\u003eNeuroimage\u003c/em\u003e \u003cstrong\u003e260\u003c/strong\u003e, 119464 (2022).\u003c/li\u003e\n\u003cli\u003eT\u0026ouml;ger, J. \u003cem\u003eet al.\u003c/em\u003e Real-time imaging of respiratory effects on cerebrospinal fluid flow in\u0026nbsp;small diameter passageways. \u003cem\u003eMagn. Reson. Med.\u003c/em\u003e \u003cstrong\u003e88\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003cli\u003eWilliams, S. D. \u003cem\u003eet al.\u003c/em\u003e Neural activity induced by sensory stimulation can drive large-scale cerebrospinal fluid flow during wakefulness in humans. \u003cem\u003ePLoS Biol.\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, e3002035 (2023).\u003c/li\u003e\n\u003cli\u003eFultz, N. E. \u003cem\u003eet al.\u003c/em\u003e Coupled electrophysiological, hemodynamic, and cerebrospinal fluid oscillations in human sleep. \u003cem\u003eScience (80-. ).\u003c/em\u003e (2019) doi:10.1126/science.aax5440.\u003c/li\u003e\n\u003cli\u003eWilliamson, N. H., Komlosh, M. E., Benjamini, D. \u0026amp; Basser, P. J. Limits to flow detection in phase contrast MRI. \u003cem\u003eJ. Magn. Reson. Open\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e\u0026ndash;\u003cstrong\u003e3\u003c/strong\u003e, (2020).\u003c/li\u003e\n\u003cli\u003eHolstein-R\u0026oslash;nsbo, S. \u003cem\u003eet al.\u003c/em\u003e Glymphatic influx and clearance are accelerated by neurovascular coupling. \u003cem\u003eNat. Neurosci.\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 1042\u0026ndash;1053 (2023).\u003c/li\u003e\n\u003cli\u003eCharidimou, A. \u003cem\u003eet al.\u003c/em\u003e The Boston criteria version 2.0 for cerebral amyloid angiopathy: a multicentre, retrospective, MRI\u0026ndash;neuropathology diagnostic accuracy study. \u003cem\u003eLancet Neurol.\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003cli\u003eDreha-Kulaczewski, S. \u003cem\u003eet al.\u003c/em\u003e Inspiration is the major regulator of human CSF flow. \u003cem\u003eJ. Neurosci.\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e, (2015).\u003c/li\u003e\n\u003cli\u003eMunting, L. P. \u003cem\u003eet al.\u003c/em\u003e Spontaneous vasomotion propagates along pial arterioles in the awake mouse brain like stimulus-evoked vascular reactivity. \u003cem\u003eJ. Cereb. Blood Flow Metab.\u003c/em\u003e (2023).\u003c/li\u003e\n\u003cli\u003eHelakari, H. \u003cem\u003eet al.\u003c/em\u003e Human NREM Sleep Promotes Brain-Wide Vasomotor and Respiratory Pulsations. \u003cem\u003eJ. Neurosci.\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003cli\u003ePerosa, V. \u003cem\u003eet al.\u003c/em\u003e Perivascular space dilation is associated with vascular amyloid-\u0026beta; accumulation in the overlying cortex. \u003cem\u003eActa Neuropathol.\u003c/em\u003e \u003cstrong\u003e143\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003cli\u003eEide, P. K. \u0026amp; Ringstad, G. Functional analysis of the human perivascular subarachnoid space. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, (2024).\u003c/li\u003e\n\u003cli\u003eHarrison, I. F. \u003cem\u003eet al.\u003c/em\u003e Non-invasive imaging of CSF-mediated brain clearance pathways via assessment of perivascular fluid movement with diffusion tensor MRI. \u003cem\u003eElife\u003c/em\u003e (2018) doi:10.7554/eLife.34028.\u003c/li\u003e\n\u003cli\u003eHelenius, J. \u003cem\u003eet al.\u003c/em\u003e Diffusion-weighted MR imaging in normal human brains in various age groups. \u003cem\u003eAm. J. Neuroradiol.\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, (2002).\u003c/li\u003e\n\u003cli\u003eBazin, P. L. \u003cem\u003eet al.\u003c/em\u003e Sharpness in motion corrected quantitative imaging at 7T. \u003cem\u003eNeuroimage\u003c/em\u003e \u003cstrong\u003e222\u003c/strong\u003e, (2020).\u003c/li\u003e\n\u003cli\u003eAndersen, M., Bj\u0026ouml;rkman-Burtscher, I. M., Marsman, A., Petersen, E. T. \u0026amp; Boer, V. O. Improvement in diagnostic quality of structural and angiographic MRI of the brain using motion correction with interleaved, volumetric navigators. \u003cem\u003ePLoS One\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, (2019).\u003c/li\u003e\n\u003cli\u003ePeper, E. S. \u003cem\u003eet al.\u003c/em\u003e Highly accelerated 4D flow cardiovascular magnetic resonance using a pseudo-spiral Cartesian acquisition and compressed sensing reconstruction for carotid flow and wall shear stress. \u003cem\u003eJ. Cardiovasc. Magn. Reson.\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, (2020).\u003c/li\u003e\n\u003cli\u003eUecker, M., Tamir, J. I., Ong, F. \u0026amp; Lustig, M. The BART Toolbox for Computational Magnetic Resonance Imaging. \u003cem\u003eIsmrm\u003c/em\u003e (2016).\u003c/li\u003e\n\u003cli\u003eHirschler, L. \u003cem\u003eet al.\u003c/em\u003e The driving force of glymphatics: influence of the cardiac cycle on CSF mobility in perivascular spaces in humans. in \u003cem\u003eProceedings of the 29th Annual Meeting of ISMRM, Sydney, Australia, 2020. Abstract 2127.\u003c/em\u003e\u003c/li\u003e\n\u003cli\u003eHirschler, L., Runderkamp, B., van Veluw, S. J., Caan, M. W. A. \u0026amp; van Osch, M. J. P. Effects of the cardiac and respiratory cycles on CSF-mobility in human subarachnoid and perivascular spaces. in \u003cem\u003eProceedings of the 31st Annual Meeting of ISMRM, London, UK, 2022. Abstract 0320.\u003c/em\u003e\u003c/li\u003e\n\u003cli\u003eKlein, S., Staring, M., Murphy, K., Viergever, M. A. \u0026amp; Pluim, J. P. W. Elastix: a toolbox for intensity based medical image registration. \u003cem\u003eIEEE Trans. Med. Imaging\u003c/em\u003e \u003cstrong\u003e29\u003c/strong\u003e, 196\u0026ndash;205 (2010).\u003c/li\u003e\n\u003cli\u003eKroon, D.-J. DTI and Fiber Tracking. https://www.mathworks.com/matlabcentral/fileexchange/21130-dti-and-fiber-tracking (2008).\u003c/li\u003e\n\u003cli\u003eMoreland, K. \u003cem\u003eet al.\u003c/em\u003e The ParaView Guide. \u003cem\u003eSandia Natl. Lab.\u003c/em\u003e (2016).\u003c/li\u003e\n\u003cli\u003eJenkinson, M., Bannister, P., Brady, M. \u0026amp; Smith, S. Improved optimization for the robust and accurate linear registration and motion correction of brain images. \u003cem\u003eNeuroimage\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, (2002).\u003c/li\u003e\n\u003cli\u003eSmith, S. M. Fast robust automated brain extraction. \u003cem\u003eHum. Brain Mapp.\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, (2002).\u003c/li\u003e\n\u003cli\u003eBeckmann, C. F. \u0026amp; Smith, S. M. Probabilistic Independent Component Analysis for Functional Magnetic Resonance Imaging. \u003cem\u003eIEEE Trans. Med. Imaging\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, (2004).\u003c/li\u003e\n\u003cli\u003eWoolrich, M. W., Ripley, B. D., Brady, M. \u0026amp; Smith, S. M. Temporal autocorrelation in univariate linear modeling of FMRI data. \u003cem\u003eNeuroimage\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, (2001).\u003c/li\u003e\n\u003cli\u003eGreve, D. N. \u0026amp; Fischl, B. Accurate and robust brain image alignment using boundary-based registration. \u003cem\u003eNeuroimage\u003c/em\u003e \u003cstrong\u003e48\u003c/strong\u003e, (2009).\u003c/li\u003e\n\u003cli\u003eJenkinson, M. \u0026amp; Smith, S. A global optimisation method for robust affine registration of brain images. \u003cem\u003eMed. Image Anal.\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, (2001).\u003c/li\u003e\n\u003cli\u003eAndersson, J. L. R., Jenkinson, M. \u0026amp; Smith, S. \u003cem\u003eNon-linear registration aka spatial normalisation\u003c/em\u003e. \u003cem\u003eFMRIB Technical Report TRO7JA2\u003c/em\u003e (2007).\u003c/li\u003e\n\u003cli\u003eFischl, B. FreeSurfer. \u003cem\u003eNeuroImage\u003c/em\u003e vol. 62 (2012).\u003c/li\u003e\n\u003cli\u003eZhang, Y., Brady, M. \u0026amp; Smith, S. Segmentation of brain MR images through a hidden Markov random field model and the expectation-maximization algorithm. \u003cem\u003eIEEE Trans. Med. Imaging\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, (2001).\u003c/li\u003e\n\u003cli\u003eYushkevich, P. A. \u003cem\u003eet al.\u003c/em\u003e User-guided 3D active contour segmentation of anatomical structures: Significantly improved efficiency and reliability. \u003cem\u003eNeuroimage\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 1116\u0026ndash;1128 (2006).\u003c/li\u003e\n\u003cli\u003eManniesing, R. \u0026amp; Niessen, W. Multiscale vessel enhancing diffusion in CT angiography noise filtering. in \u003cem\u003eLecture Notes in Computer Science\u003c/em\u003e vol. 3565 (2005).\u003c/li\u003e\n\u003cli\u003eMeijering, E. \u003cem\u003eet al.\u003c/em\u003e Neurite tracing in fluorescence microscopy images using ridge filtering and graph searching: Principles and validation. in \u003cem\u003e2004 2nd IEEE International Symposium on Biomedical Imaging: Macro to Nano\u003c/em\u003e vol. 2 (2004).\u003c/li\u003e\n\u003cli\u003eTeam, J. JASP Team (Version 0.17). (2023).\u003c/li\u003e\n\u003cli\u003eHirschler, L. \u003cem\u003eet al.\u003c/em\u003e High resolution T2-prepared MRI enables non-invasive assessment of CSF flow in perivascular spaces of the human brain. in \u003cem\u003eProceedings of the 28th Annual Meeting of ISMRM, Montr\u0026eacute;al, Canada, 2019. Abstract 0746.\u003c/em\u003e\u003c/li\u003e\n\u003cli\u003eVan Osch, M. J. P., Petitclerc, L. \u0026amp; Hirschler, L. Probing cerebrospinal fluid mobility for human brain clearance imaging MRI: water transport across the blood-cerebrospinal fluid barrier and mobility of cerebrospinal fluid in perivascular spaces. \u003cem\u003eVeins Lymphat.\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-3178346/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3178346/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eHighlights\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eCSF-mobility can be measured in humans down to the level of perivascular spaces surrounding penetrating vessels using CSF-STREAM\u003c/li\u003e\n \u003cli\u003eIn perivascular spaces surrounding penetrating vessels, the cardiac and respiratory cycles have similar effects on CSF-mobility\u003c/li\u003e\n \u003cli\u003eEntraining vasomotion at 0.1 Hz can drive CSF-mobility in humans\u003c/li\u003e\n \u003cli\u003eCSF-mobility is altered in patients with cerebral amyloid angiopathy\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u003cstrong\u003eAbstract\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMany neurological diseases are characterized by the accumulation of toxic proteins in the brain. This accumulation has been associated with improper clearance from the parenchyma. Recent discoveries highlighted perivascular spaces, which are cerebrospinal fluid (CSF)-filled spaces, as the channels of brain clearance. The forces driving CSF-mobility within perivascular spaces are still debated. Here, we present a new, non-invasive, CSF-specific magnetic resonance imaging technique (CSF-STREAM), that enables detailed in vivo measurement of CSF-mobility in humans, for the first time down to the level of perivascular spaces located around penetrating vessels, i.e. close to protein production sites. We find region-specific drivers for CSF-mobility and demonstrate that CSF-mobility can be increased by entraining vasomotion. Furthermore, we found region-specific CSF-mobility alterations in patients with cerebral amyloid angiopathy, a brain disorder associated with clearance impairment. The availability of this new technique opens up avenues to investigate the impact of CSF-mediated clearance in neurodegeneration and sleep.\u003c/p\u003e","manuscriptTitle":"Region specific drivers of cerebrospinal fluid mobility as measured by high-resolution non-invasive MRI in humans","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-10-18 04:52:33","doi":"10.21203/rs.3.rs-3178346/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-neuroscience","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"neuro","sideBox":"Learn more about [Nature Neuroscience](http://www.nature.com/neuro/)","snPcode":"","submissionUrl":"","title":"Nature Neuroscience","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Research","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"157b48ba-c33e-4e66-a869-f7cdd47a268d","owner":[],"postedDate":"October 18th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":36312952,"name":"Biological sciences/Biological techniques/Imaging/Magnetic resonance imaging"},{"id":36312953,"name":"Biological sciences/Physiology/Neurophysiology"},{"id":36312954,"name":"Biological sciences/Neuroscience/Neuro\u0026#x2013;vascular interactions"}],"tags":[],"updatedAt":"2025-10-15T07:06:41+00:00","versionOfRecord":{"articleIdentity":"rs-3178346","link":"https://doi.org/10.1038/s41593-025-02073-3","journal":{"identity":"nature-neuroscience","isVorOnly":false,"title":"Nature Neuroscience"},"publishedOn":"2025-10-14 04:00:00","publishedOnDateReadable":"October 14th, 2025"},"versionCreatedAt":"2024-10-18 04:52:33","video":"","vorDoi":"10.1038/s41593-025-02073-3","vorDoiUrl":"https://doi.org/10.1038/s41593-025-02073-3","workflowStages":[]},"version":"v1","identity":"rs-3178346","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3178346","identity":"rs-3178346","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}