Imaging Biomarkers of the Glymphatic System in the Developing Brain

Abstract

The glymphatic system is increasingly recognized as a brain-wide pathway that mediates cerebrospinal fluid–interstitial fluid exchange and metabolic waste clearance. However, unlike in adults, its developmental trajectory and clinical relevance during infancy and childhood remain poorly understood. This review summarizes imaging biomarkers of glymphatic structure and function across neurodevelopment, with a particular emphasis on infancy, while also addressing the broader pediatric population. We focus on noninvasive MRI-based measures applicable to pediatric practice, including perivascular space (PVS) grading and volumetry, the diffusion tensor imaging–analysis along the perivascular space (DTI-ALPS) index, and choroid plexus (CP) volume. Developmental evidence indicates that PVS visibility and volume, the DTI-ALPS index, and CP volume undergo age-related changes during early development. However, age-specific normative data remain limited, particularly between infancy and school age. In pediatric populations, glymphatic-related imaging alterations have been reported in preterm birth, neonatal hypoxic-ischemic injury, and neurodevelopmental disorders such as autism spectrum disorder. Collectively, these findings highlight the potential clinical utility of non-contrast glymphatic imaging biomarkers in children and underscore the need for larger longitudinal and multicenter studies to establish normative reference values and clarify their prognostic significance.

INTRODUCTION

The glymphatic system, first proposed by Iliff et al. in 2012 [1], has been increasingly recognized as a brain-wide waste clearance pathway in the central nervous system [2]. It facilitates the exchange of cerebrospinal fluid (CSF) and interstitial fluid (ISF) through perivascular channels, a process mediated by aquaporin-4 water channels [3]. This system plays a critical role in the clearance of neurotoxic metabolites, including amyloid-β and tau proteins, and is well characterized in the adult brain [4]. However, its developmental trajectory and functional role in early childhood remain poorly understood. Elucidating glymphatic function during early development may provide important insights into the biological mechanisms underlying brain injury and subsequent neurodevelopment in young children [5,6].

MRI has served as the primary tool for noninvasive assessments of glymphatic function in early brain development. MRI-based measures include perivascular space (PVS) volume [7], diffusion tensor imaging–analysis along the perivascular space (DTI-ALPS) index [8], and choroid plexus (CP) volume [6], all of which have demonstrated potential clinical relevance [5,9,10,11]. In addition, ultrasound has emerged as a promising imaging modality that may be uniquely applicable during the neonatal period [12]. Although debate persists regarding whether these imaging measures directly reflect glymphatic function, accumulating evidence supports their utility as surrogate markers of glymphatic function [5,6,7]. Accordingly, in this review, we collectively refer to them as imaging biomarkers of glymphatic function, while acknowledging the limitations inherent to each measure. This review summarizes imaging biomarkers of glymphatic function across brain development, with a primary focus on infancy and extension to the broader pediatric age range, emphasizing their clinical relevance and key considerations for radiologists. We also discuss current challenges and future directions in this emerging field.

IMAGING BIOMARKERS OF GLYMPHATIC FUNCTION IN THE DEVELOPING BRAIN

Perivascular Spaces: Grading and Volume Assessment

PVSs, also known as Virchow–Robin spaces, are fluid-filled channels that surround penetrating cerebral vessels and are structurally linked to the perivascular pathway through which CSF and ISF exchange occurs (Fig. 1) [1]. The glymphatic system facilitates metabolic waste clearance and maintains brain fluid homeostasis by promoting CSF influx into the interstitial space [1,13]. Historically, PVSs were regarded as normal anatomical structures with limited clinical significance, attracting attention only when markedly dilated in diseases such as mucopolysaccharidosis [14,15]. However, increased recognition of the glymphatic system and the role of PVSs as a key component of its pathway has led to growing interest in their imaging characteristics and quantification [16]. Recent evidence suggests that PVS visibility and burden are closely associated with physiological states and glymphatic function, and that PVS development is a gradual process beginning in early life and continuing into adulthood [7].

Fig. 1. Imaging biomarkers of glymphatic function in the developing brain. This figure illustrates four distinct imaging biomarkers used to evaluate glymphatic function in neonates: 1) white matter-PVS (top), 2) basal ganglia-PVS (left), 3) the DTI-ALPS index (right), and 4) the choroid plexus (bottom). PVS = perivascular space, DTI-ALPS = diffusion tensor imaging–analysis along the perivascular space.

Given their association with glymphatic function, PVS quantification has become an important focus of neuroimaging studies, encompassing both semiquantitative approaches (e.g., visual grading) and quantitative methods (e.g., automated or semi-automated segmentation) [17]. On T2-weighted MRI, PVS appears as linear or round high-signal-intensity structures, isointense with CSF [18]. Although PVS can be observed across multiple brain regions [19], most imaging studies have focused on two major regions for semiquantitative or quantitative PVS assessment: the cerebral basal ganglia (BG) and white matter (WM) [20,21].

Visual grading is a commonly used semiquantitative method in which trained raters visually assess the number of visible PVS on axial T2-weighted images at predefined anatomic levels and assign an ordinal score reflecting PVS burden. A five-point scale is applied separately for the BG and WM: grade 0 indicates no PVS; grade 1, 1–10 PVS; grade 2, 11–20; grade 3, 21–40; and grade 4, more than 40 PVS [21,22]. These PVS are counted in either the right or left hemisphere, selecting the axial section that demonstrates the greatest number of PVS [23]. For BG assessment, PVSs at or below the level of the anterior commissure are excluded [23]. This exclusion is applied because PVSs in the BG (BG-PVS) at this level are visible in most individuals, limiting the ability of visual grading to detect clinically meaningful variability when included [21].

At term-equivalent age, BG-PVS was graded as 0 in 37% of neonates and as grade 1 in 63%, whereas PVS in the WM (WM-PVS) was graded as 0 in 93% of cases [7]. Based on this distribution, a high PVS burden above the 90th percentile has been defined as grade ≥2 for BG-PVS and grade ≥1 for WM-PVSs in neonates. Although visual grading enables standardized assessment, it is subject to inherent ceiling and floor effects.

For quantitative approaches, PVS segmentation is typically performed on three-dimensional (3D) T2-weighted images using computational methods designed to enhance tubular structures, such as Frangi filtering [24], followed by manual refinement to reduce false positives and false negatives, and improve anatomical precision [25]. More recently, anatomy-informed deep learning approaches for joint BG and PVS segmentation have been proposed to improve performance in infant brain MRI [26]. However, PVS segmentation in infants remains challenging due to their smaller size, lower signal-to-noise ratio, and reduced tissue contrast compared with adult imaging [27]. Furthermore, reliable PVS extraction in specific brain regions requires accurate regional brain segmentation, which is also complicated by the inherent neuroimaging differences between pediatric and adult brains. To address these challenges, infant-specific processing pipelines, such as Infant FreeSurfer [28] and BabySeg [29], have been developed to facilitate BG-PVS extraction. Unlike visual grading of BG-PVS, quantitative segmentation typically includes PVS at and below the level of the anterior commissure. As ultrasound remains one of the primary neuroimaging modalities in neonates, cranial ultrasound has also been explored for the assessment of PVS. A prior study demonstrated that PVSs in the caudal and lateral regions of the BG correspond to hyperechogenic structures on neonatal cranial ultrasound, which can be quantitatively assessed using 3D measurements derived from sagittal and coronal ultrasound images [12]. Although cranial ultrasound-derived PVS volume shows good correlation with MRI-derived measurements, the PVS regions assessed by ultrasound (near the origin of the lenticulostriate vessels at the caudal and lateral portions of the BG) differ from those typically evaluated in previous MRI studies, warranting further validation.

Perivascular Spaces: DTI-ALPS Index

The DTI-ALPS index is a diffusion-based imaging biomarker designed to assess glymphatic function or ISF dynamics in the brain. Unlike the structural information provided by PVS volumetry, the DTI-ALPS index has been proposed as a measure of the functional aspects of the glymphatic system. Specifically, the DTI-ALPS index quantifies water diffusivity along PVS using DTI (Fig. 1) [30]. Diffusivities along the x-, y-, and z-axes are measured within projection, association, and fiber tracts, respectively. Regions of interest (ROIs) are placed at the level of the lateral ventricle body. The DTI-ALPS index is calculated as the ratio of diffusivity along the PVS to diffusivity along the dominant fiber directions, providing an indirect measure of water diffusivity along PVS pathways [30].

Compared with automated PVS segmentation, the DTI-ALPS method is computationally less demanding because it relies on relatively simple calculations derived from DTI, which is commonly acquired in routine pediatric neuroimaging protocols [30,31]. However, the method is highly operator-dependent, particularly with respect to axial slice selection and ROI placement [32]. These challenges are further amplified in pediatric populations, particularly neonates, because of their smaller brain size and less distinct anatomical landmarks. To accommodate neonatal brain anatomy, ROI diameters are typically reduced from the standard adult size of 5 mm to approximately 3–4 mm [5,6]. Furthermore, precise anatomical localization and standardized head positioning are critical for reliable DTI-ALPS measurements [33], yet achieving these requirements can be challenging in young children. Consequently, interobserver variability remains a significant limitation of DTI-ALPS assessment during early brain development and should be carefully considered when interpreting findings in pediatric populations [34].

Choroid Plexus Volume

The CP is responsible for CSF production and plays a central role in maintaining intracranial fluid homeostasis (Fig. 1) [35]. Within the glymphatic system, the CP regulates CSF turnover, a process essential for driving the convective clearance of metabolic waste through the perivascular network [36]. Historically, the CP was regarded primarily as a CSF-producing structure and, similar to PVS, received limited attention in neuroimaging studies. In routine clinical practice, CP assessment has largely been limited to cases with overt pathology, such as CP tumors, cysts, or hemorrhage. More recently, however, CP volume has been proposed as a surrogate biomarker of glymphatic function, as it may reflect CSF secretory capacity [37,38].

Unlike PVS, no standardized visual grading system currently exists for assessing CP size. Consequently, most studies quantify CP volume on structural MRI using either manual segmentation [39] or automated approaches [40], including Gaussian mixture model-based approaches [41,42] and deep learning models, such as optimized 3D U-Net approaches [43,44]. A lifespan-scale study further applied pipelines combining automated segmentation with expert annotation and refinement (human-in-the-loop framework) [45]. CP segmentation is most commonly performed on high-resolution 3D T1-weighted images [41,46], which provides superior contrast between ventricular CSF and CP tissue. To enhance boundary delineation, some protocols incorporate additional sequences, including T2-weighted and T2-weighted fluid-attenuated inversion recovery (FLAIR) images [47]. Although contrast-enhanced T1-weighted imaging may serve as the reference standard, non-contrast T1-weighted imaging has also been reported to provide an acceptable surrogate for CP volumetry, whereas T2-weighted FLAIR imaging has been shown to overestimate CP volume [46]. Nevertheless, a recent study using deep learning approaches demonstrated comparable CP segmentation performance across non-contrast T1-weighted, T2-weighted, and T2-weighted FLAIR sequences [43]. Table 1 provides a summary of the three glymphatic imaging markers in the developing brain.

Table 1. Key MRI-based imaging biomarkers related to glymphatic function in developing brain.

Target anatomy	Biomarker	MRI sequence	Description	Key limitations
PVS	Visual grading	2D T2-weighted images; 3D T2-weighted images	Semi-quantitative grading (grade 0–4) based on the number of visible PVS on axial images	Rater dependence; limited sensitivity
	Volume	3D T2-weighted images	Quantification of PVS volume using region segmentation	Sensitive to image quality and sequence parameters
	DTI-ALPS index	DTI	Measurement of water diffusivity along PVS in cerebral WM	Operator-dependent ROI and slice selection Influenced by WM maturation and fiber geometry
CP	Volume	3D T1-weighted images	Quantification of CP volume using region segmentation	Sensitive to image quality and sequence parameters

PVS = perivascular space, D = dimensional, DTI-ALPS = diffusion tensor imaging–analysis along the perivascular space, WM = white matter, ROI = region of interest, CP = choroid plexus

DEVELOPMENTAL CHANGES OF GLYMPHATIC IMAGING BIOMARKERS

Studies have demonstrated that PVS undergoes dynamic changes in visibility and volume throughout early brain development [7,48]. In neonates, the BG-PVS fraction, defined as PVS volume normalized to BG volume, decreases with increasing postmenstrual age at MRI [7]. It is important to note that, unlike in older children and adults, age in neonatal imaging is often described using postmenstrual age (also referred to as corrected gestational age), which incorporates both gestational age at birth and postnatal age. In contrast, absolute PVS volume in neonates does not appear to change significantly with age. WM-PVS volume was not evaluated in the same study because visible WM-PVS were absent in most cases [7]. In a normative cohort spanning a broader age range, both WM-PVS and BG-PVS volumes were negatively associated with age from 8 years through the early to mid-30s [48], suggesting a continued reduction in PVS volume throughout childhood beyond 8 years of age. However, data characterizing PVS volume changes between the neonatal period and 8 years of age remain limited, highlighting the need for validation in large, population-based normative cohorts.

Developmental changes in the DTI-ALPS index have also been investigated in the developing brain [49]. In a study of 418 typically developing neonates from the Developing Human Connectome Project, the DTI-ALPS index was positively correlated with postmenstrual age [8]. Similarly, a longitudinal study of children and adolescents reported an age-related increase in the DTI-ALPS index [49]. However, this positive association should be interpreted with caution, as the DTI-ALPS index may be influenced by WM development itself rather than solely reflecting water diffusivity along the PVS [32]. DTI metrics, including fractional anisotropy and mean diffusivity, undergo substantial changes during myelination and may consequently influence the DTI-ALPS index [50]. Furthermore, recent evidence indicates that WM geometry, including crossing fibers, axonal undulations, and orientation dispersion, can confound DTI-ALPS-derived measures [32].

CP volume in neonates is positively associated with postmenstrual age at the time of imaging [6]. This association remains significant after adjustment for potential confounders, suggesting that CP growth is linked to the maturational regulation of CSF production during the perinatal period [6]. CT–based studies of children aged 0–10 years have shown rapid CP growth during the first year of life, followed by a plateau at approximately 1.5 mL [51]. In contrast, an MRI-based volumetric study reported that CP volume remains relatively stable during the first 2 years of life and subsequently increases progressively throughout childhood and adolescence [45].

Figure 2 presents a summary of the developmental trajectories of three MRI-based glymphatic-related biomarkers from birth to 18 years of age. These trajectories should be interpreted as approximate, as they are derived from a limited number of studies with heterogeneous methodologies, age distributions, and sample sizes.

Fig. 2. Developmental changes in glymphatic imaging biomarkers. PVS volume (orange) declines continuously from birth to 18 years of age. In neonates, the PVS fraction decreases with increasing gestational and postmenstrual age [7], and this pattern appears to extend into later childhood and adolescence, consistent with a negative association between PVS volume and age in individuals aged 8–18 years [20]. CP volume (blue) remains relatively stable during the first 2 years of life and then increases approximately linearly throughout childhood and adolescence [45]. The DTI-ALPS index (green) increases continuously from birth through adolescence; in neonates, it is positively associated with gestational and postmenstrual age [8], and this increasing trend persists during childhood and adolescence [49]. All metrics are reported in a.u. for comparative purposes. Dashed lines indicate extrapolated trajectories for ages 1–6 years, during which empirical data are limited. It should be noted that these developmental trajectories are simplified representations based on a limited number of studies with heterogeneous methodologies and sample sizes and should be interpreted with caution. PVS = perivascular space, CP = choroid plexus, DTI-ALPS = diffusion tensor imaging–analysis along the perivascular space, a.u. = arbitrary units.

GLYMPHATIC IMAGING BIOMARKERS AND PEDIATRIC DISORDERS

Preterm Birth

Preterm birth, defined as delivery before 37 weeks of gestation, is a major global health concern associated with increased risks of mortality, neurodevelopmental impairment, and long-term medical complications. Preterm birth has also been associated with alterations in the glymphatic system. Regarding PVS, one study reported lower BG-PVS visual grading scores and lower BG-PVS volume in preterm neonates compared with term neonates (Fig. 3) [7]. In contrast, a more recent study found no difference in BG-PVS counts but reported higher WM-PVS counts in preterm neonates [52]. Similarly, the DTI-ALPS index was lower in preterm than in term neonates [8]. With respect to CP volume, preterm neonates exhibited smaller CP volumes than term neonates; however, the difference was not statistically significant after adjustment for potential confounders [6].

Fig. 3. Associations of gestational age, sex, and HII with glymphatic biomarkers in neonates. A: CP volume differences by gestational age (left) and sex (right) [6]. Preterm neonates initially exhibited smaller CP volumes than term neonates. However, this difference was no longer statistically significant after adjustment for covariates, suggesting that the unadjusted association likely reflects maturation-related effects rather than prematurity per se. In addition, biological sex was not a significant determinant of neonatal CP volume. Reprinted from Peng et al., BMC Med Imaging 2025;25:126 [6], under a CC BY-NC-ND license. B: Glymphatic alterations associated with HII [5]. Neonates with HII demonstrated significantly lower BG-PVS volume (left) and fraction (middle) than those without HII, whereas the DTI-ALPS index (right) did not differ between groups. Reprinted from Choi et al., Korean J Radiol 2025;26:782-792 [5], under a CC BY-NC license. HII = hypoxic-ischemic injury, CP = choroid plexus, BG-PVS = basal ganglia perivascular space, DTI-ALPS = diffusion tensor imaging–analysis along the perivascular space, ns = not significant.

Hypoxic-Ischemic Injury

Neonatal hypoxic-ischemic injury (HII) is a major cause of acute brain injury in neonates and is associated with long-term neurodevelopmental impairment. One study reported significantly smaller BG-PVS volumes and fractions in neonates with HII than in healthy controls [5]. In the same study, no significant difference in the DTI-ALPS index was observed between neonates with and without HII (Fig. 3) [5]. In contrast, another study reported a lower DTI-ALPS index in neonates with HII than in controls [11]. Although the CP is particularly vulnerable to HII in the immature brain [53], studies investigating CP volume in neonatal HII remain scarce. Given the role of CP in post-injury inflammatory responses and CSF dynamics [54], further research is needed to clarify CP alterations in neonatal HII.

Neurodevelopmental Disorders

Neurodevelopmental disorders comprise a heterogeneous group of conditions characterized by impairments in cognitive, behavioral, motor, or social functioning arising during early brain development. Among these disorders, autism spectrum disorder (ASD) is one of the most extensively studied in relation to glymphatic function. A prospective longitudinal cohort study demonstrated that infants who were later diagnosed with ASD had a higher prevalence of enlarged PVS than infants who were not diagnosed with ASD [10]. Similarly, a pilot study reported that increased BG-PVS volume in neonates was associated with poorer cognitive and motor performance at 24 months [9]. Regarding the DTI-ALPS index, children with ASD or attention-deficit/hyperactivity disorder exhibited a lower DTI-ALPS index compared with healthy controls [55,56,57]. Regarding CP volume, one study reported increased CP volume in adults with ASD [58]; however, data on CP volume in pediatric patients with ASD remain limited.

Other Pediatric Neurological and Non-Neurological Disorders

We did not elaborate further on detailed findings from other neurological disorders that may influence glymphatic structures in the developing brain. Nevertheless, accumulating evidence suggests that several conditions, including epilepsy [59,60], hydrocephalus [61,62], and primary headache [63,64], are associated with alterations in glymphatic parameters assessed on brain MRI. In pediatric populations, typically older than 2 years of age, studies have consistently reported a greater PVS burden [59,63,64] and lower DTI-ALPS indices [55,56,57,60,61] in various neurologic conditions compared with healthy controls. Similar findings have been reported in non-neurological conditions. For example, studies of pediatric hematologic malignancies have demonstrated increased PVS burden and reduced DTI-ALPS indices in affected patients [65]. In contrast, data regarding CP volume in pediatric populations remain limited. Table 2 summarizes the published literature on glymphatic parameters across a range of pediatric diseases [5,6,7,8,9,10,11,55,56,57,59,60,61,62,64,65].

Table 2. Summary of glymphatic system imaging parameters across pediatric conditions.

Pediatric disorders	PVS volume	DTI-ALPS index	CP volume
Preterm birth	↓ [7]	↓ [8]	↔ * [6]
Hypoxic-ischemic injury	↓ [5]	↔ [5], ↓ [11]	-
Neurodevelopmental disorders†	↑ [9,10]	↓ [55,56,57]	-
Epilepsy	↑ [59]	↓ [60]	-
Hydrocephalus	-	↓ [61]	↑ [62]
Headache	↑ [64]	-	-
Hematologic disease	-	↓ [65]	-

^*In preterm versus term neonates, CP volume was significantly lower before adjustment for confounders but not significant after adjustment, ^†Neurodevelopmental disorders include autism spectrum disorder and attention-deficit/hyperactivity disorder. ↑ indicates significantly increased volume, visibility, or index compared to controls; ↓ indicates significantly decreased volume, visibility, or index compared to controls; ↔ indicates no significant difference; - indicates insufficient data or no specific pediatric studies available.

CP = choroid plexus, PVS = perivascular spaces, DTI-ALPS = diffusion tensor imaging–analysis along the perivascular space

CHALLENGES IN GLYMPHATIC IMAGING IN THE DEVELOPING BRAIN

In addition to the three glymphatic imaging markers discussed in this review, several advanced imaging techniques have been developed to directly visualize and quantify glymphatic function, many of which require the administration of contrast agents. These approaches include dynamic contrast-enhanced MRI for assessing meningeal lymphatic flow [66] and intrathecal contrast-enhanced MRI [67]. Although these techniques have been successfully applied in adult populations, their use in pediatric research is limited by practical and ethical considerations, underscoring the need to further refine non-contrast-enhanced imaging biomarkers for children. Technical challenges also complicate glymphatic imaging in pediatric populations. Image quality is often degraded by patient motion during scanning and by the relatively small size of the pediatric brain. Although sedation may reduce motion artifacts, its use raises concerns regarding both short- and long-term adverse effects. At the same time, many children are unable to remain motionless during MRI examinations without sedation, resulting in image degradation. Motion artifacts can impair image processing and reduce the accuracy of segmenting small anatomical structures, including PVS and the CP.

The glymphatic imaging markers reviewed here also have important limitations. Although differences between patient and control groups have been reported, pathological validation in children remains lacking. This contrasts with adult studies, in which postmortem investigations have provided histological confirmation of glymphatic imaging findings [68]. Enlarged PVS may result not only from decreased CSF-ISF exchange, but also from brain atrophy and other pathological conditions such as neuroinflammation. Therefore, careful interpretation of enlarged PVS as a glymphatic biomarker is warranted. Furthermore, glymphatic imaging markers are highly sensitive to imaging acquisition parameters and post-processing methodologies, including sequence selection, spatial resolution, post-processing thresholds, and ROI placement. Variations in these factors can substantially influence biomarker measurements. This issue is particularly relevant for the DTI-ALPS index, which is affected by developmental changes in WM diffusivity. As discussed earlier, conditions associated with reduced WM anisotropy may lower the DTI-ALPS index because of increased radial diffusivity within WM rather than altered water diffusivity in the PVS [32,34]. Therefore, cross-age comparisons should be interpreted with caution and should account for age-related microstructural maturation and changes in WM DTI metrics. Collectively, these limitations highlight the need for careful interpretation of glymphatic imaging findings and emphasize the importance of methodological standardization in pediatric glymphatic imaging.

CONCLUSION AND FUTURE DIRECTIONS

This review summarizes current evidence regarding noninvasive MRI-based biomarkers for assessing glymphatic function in the developing brain, including PVS grading and volume, the DTI-ALPS index, and CP volume. These biomarkers exhibit dynamic developmental changes throughout childhood and are altered in a range of pediatric conditions, including preterm birth, HII, and neurodevelopmental disorders such as ASD. These findings suggest their potential utility for the early identification of children at risk for adverse neurological outcomes. However, normative developmental data remain limited and fragmented, with substantial knowledge gaps, particularly during the transitional period between infancy and school age. Standardized imaging acquisition protocols, age-specific reference ranges, and longitudinal validation studies are needed to facilitate clinical translation. Future large-scale, multicenter investigations are essential to establish robust normative datasets and to clarify the prognostic significance of glymphatic imaging biomarkers in pediatric populations.

Availability of Data and Material

All data generated or analyzed during this study are included in this published article.