As previously mentioned, conventional T2-weighted and FLAIR sequences remain the standard for WMH delineation (Figure 1A) [33,39-41]. These contrasts are highly sensitive to increased tissue water: damaged white matter exhibits prolonged T2 relaxation, appearing hyperintense. FLAIR improves periventricular lesion detection by suppressing cerebrospinal fluid (CSF) signal. Accordingly, WMHs are defined as hyperintense regions identified through intensity-based methods on T2-weighted (including FLAIR) MRI, predominantly located in the periventricular and deep white matter.

Although WMHs were traditionally evaluated visually—with lesion burden often graded qualitatively using scales such as the Fazekas score, or delineated manually for volumetric estimation [24,42-44]—recent methodological advances have led to the widespread adoption of semi-automated and fully automated approaches, enabling more detailed and quantitative analyses [45-47]. A potential challenge arises in stroke patients when there is a delay between stroke onset and FLAIR acquisition: acute ischemic lesions, which appear hyperintense on diffusion-weighted imaging (DWI), may also exhibit hyperintensity on T2-weighted or FLAIR sequences, complicating their distinction from WMH [48-50]. In such instances, the typically symmetric distribution of WMH can assist in differentiation [33,51,52]. Nevertheless, caution is warranted, as asymmetric WMH patterns may occur and can themselves convey additional clinically relevant information [53].

Historically, WMHs were categorized into PVWMH and DWMH subtypes based on visual assessment of their spatial relationship to the lateral ventricles [3,23-25,42,54]. The principal criterion was continuity with the ventricular wall: PVWMH were defined by direct contact with the lateral ventricular wall, whereas DWMH were spatially separate. In practice, this rule proved fragile. With WMH progression, confluent PVWMH often merges with DWMH, obscuring boundaries. Inconsistent use of ancillary features (e.g., lesion shape), arbitrary distance thresholds, and divergent reassignment rules for large or irregular lesions further undermined reproducibility, likely contributing to cross-study heterogeneity in reported prevalence, risk factors, and clinical associations [20,42]. To enhance objectivity—even if modestly—contemporary schemes adopt explicit distance-based taxonomies measured from the ventricular surface. An extended framework [20,54-57] further links spatial location to putative mechanistic distinctions, thereby reducing inter-study variability. It delineates four territories: juxtaventricular WMH (JVWMH), contiguous with the ependyma within ~0-3 mm of the ventricle; PVWMH, occupying the periventricular “border-zone” at ~3-13 mm; DWMH, located at ≥~13 mm; and juxtacortical WMH (JCWMH), within ~4 mm of the corticomedullary junction. This distance-informed classification reduces ambiguity and provides a spatial framework that facilitates interpretation of lesion location in relation to biologically relevant patterns.

Postmortem histopathology indicates that WMH lesions include axonal loss and demyelination 16,18,19,58]; these changes reduce fiber coherence and expand the extracellular space, thereby altering water diffusion and tissue water content (H₂O) [59].

Diffusion tensor imaging (DTI) quantifies directional water diffusion along white matter tracts by modeling diffusion anisotropy [60,61]. It is derived from DWI, which measures the mobility of water molecules constrained by tissue microstructure [62,63]. In intact tissue, axons and myelin restrict diffusion—particularly perpendicular to fibers—whereas in WMH, the loss of these microstructural barriers permits more isotropic diffusion of water molecules [59].

Standard DTI metrics include fractional anisotropy (FA), reflecting directional coherence; mean diffusivity (MD), reflecting overall diffusivity; axial diffusivity (AD), diffusion along the principal axis; and radial diffusivity (RD), diffusion perpendicular to fibers (Figure 1B). These metrics have been histologically validated as markers of white matter microstructure, primarily capturing aspects of axonal organization rather than myelin den-sity [64-66]. Compared with NAWM, WMHs consistently shows lower FA [27,67-72] and higher MD [27,67-69,71,72] and RD [67,73], consistent with reduced fiber coherence and extracellular space expansion [16].

Notably, and contrary to expectation, AD is often higher in WMH than in NAWM [67,69,73]. Because AD reflects diffusion along the principal fiber axis, this elevation does not indicate preserved axons. Rather, it likely reflects tissue rarefaction—including vasogenic edema, myelin and cellular loss, and matrix disorganization—that reduces diffusion barriers, and elevates MD and RD and, sometimes, AD [74,75]. By contrast, AD decreases, when present, likely indicate advanced, chronic axonal disruption and collapse [76,77].

Diffusion kurtosis imaging (DKI) extends DTI by capturing non-Gaussian diffusion arising from tissue complexity, with summary measures such as mean kurtosis (Mk), axial kurtosis (Ak), and radial kurtosis (Rk) [78,79]. Recent postmortem studies confirmed histological correlations between DKI metrics and demyelination, both in humans [80] and in rodent models [81-83].

More advanced models further estimate the free water (FW) fraction, which is typically elevated in and around WMH [70,84] and corresponds to the characteristic FA↓/MD↑/RD↑ profile, indicating extracellular expansion and edema. However, direct histological validation of FW fraction remains limited because fixation and postmortem shrinkage alter extracellular geometry and remove FW [85].

Intravoxel incoherent motion (IVIM) imaging based on multib-value DWIs separates true parenchymal diffusion (D) from microvascular pseudo-diffusion (D*), yielding three parameters: D coefficient, D* coefficient, and the perfusion fraction (f), the latter reflecting relative microvascular volume (Figure 1C) [86,87]. This separation is possible because microvascular perfusion contributes predominantly to the diffusion signal at low b-values but diminishes at higher b-values, allowing model-based quantification of perfusion-related effects that would otherwise confound tissue water diffusion measurements. IVIM metrics have been validated in both phantom- and histology-based studies to effectively disentangle diffusion and perfusion components [88,89].

IVIM-derived D—minimizing vascular contamination—is generally higher in WMH relative to NAWM [87,90,91], providing information comparable to DTI-based metrics. A small pilot study (n=19), however, reported lower D values in WMH, possibly reflecting neovascularization, although the limited sample size precludes firm inference [92]. f is also typically higher in WMH than in NAWM [87,90,92], likely reflecting compensatory microcirculatory vasodilation that maintains oxygen delivery following white matter ischemia [90,93,94]. However, a recent study has reported lower f in WMH [84], underscoring methodological variability and the need for standardized analysis. Findings for D* remain inconsistent across studies [87,90,92], likely due to differences in acquisition protocols, lesion topography, and NAWM definitions (e.g., inclusion or exclusion of perilesional tissue). A decrease in D* may be biologically plausible—consistent with reduced cerebral blood flow (CBF) from vascular rarefaction, arteriolar tortuosity, or venous collagen deposition [95]—yet a report of increased D* highlights the parameter’s limited robustness (low signal-to-noise ratio) and model sensitivity, warranting cautious interpretation [87].

Single-shell 3-tissue constrained spherical deconvolution, also derived from DWI, enables voxelwise characterization of WMHs by decomposing the diffusion signal into three tissue-like signal fractions—white matter-like (T_w), gray matter-like (T_g), and CSF-like (T_c)—further providing a compositional description of microstructural change [96]. These tissue-like compartments do not directly represent pathology but indicate the degree of similarity to healthy tissue profiles; an increased CSF-like fraction may reflect interstitial fluid expansion or rarefaction, whereas an increased T_g may indicate gliosis within white matter. Multi-tissue diffusion modeling further demonstrates a shift toward T_c and away from T_w within WMH, underscoring a compositional transition toward fluid-rich, structurally degraded tissue [55,97].

Beyond microstructural and diffusional alterations, water mobility along perivascular spaces (PVS) provides additional insight into glymphatic clearance dysfunction, another key mechanism implicated in WMH pathology. The DTI analysis along the perivascular space (DTI-ALPS) index quantifies water diffusivity along these spaces, with lower values suggesting impaired glymphatic clearance associated with WMH [98-100]. This index is calculated in regions adjacent to the lateral ventricles, where medullary veins—and their accompanying PVS—run perpendicular to major projection and association fibers. This geometric arrangement allows estimation of water diffusivity that predominantly reflects perivascular flow with minimal contamination from surrounding white matter tracts. However, interpretation of the DTIALPS index as a direct measure of glymphatic function should be approached with caution, as it primarily quantifies water motion rather than directly capturing the complex dynamics of interstitial fluid clearance [101]. Therefore, complementary methods are required for a more comprehensive evaluation of the glymphatic system. For example, enlarged PVS [102,103]—CSF-filled cavities best visualized on T2/FLAIR MRI [104]—are frequently contiguous with or adjacent to WMH [105,106] and are considered structural correlates of impaired CSF clearance through the glymphatic pathway [107-109]. Their enlargement may signify impaired interstitial fluid drainage and reduced glymphatic clearance.

Whereas DWI provides general indices of water diffusion and H₂O, quantitative MRI (qMRI) techniques aim to quantify myelin and related macromolecules more specifically [16,66,110]. qMRI complements structural imaging with measures sensitive to specific tissue properties. qMRI extends conventional structural imaging by providing calibrated, biologically meaningful measurements sensitive to myelination, iron content, and cell membrane integrity in the living brain. A range of derived techniques—including relaxation rate mapping, myelin partial volume (V_MY) and H₂O mapping, T1w/T2w ratio, myelin water imaging (MWI), and magnetization transfer imaging (MTI)—are collectively referred to here as qMRI approaches for in vivo characterization of macromolecular integrity. The longitudinal relaxation rate R1 (1/T1) is higher in myelin-rich tissue, whereas the effective transverse relaxation rate R2* (1/T2*) is more influenced by iron content (Figure 1A) [111-116]. Additional parameters include V_MY—a more direct estimate of myelin content—and H₂O, which indexes edema [117]. Together, R1, R2*, V_MY, and H₂O constitute a practical panel for probing tissue integrity. Across studies, WMHs compared with NAWM consistently show lower R1 (1/T1) [27,57,118], R2* (1/T2*) [57,118], and V_MY [119], along with higher H₂O [118], a pattern indicative of demyelination and tissue rarefaction. The T1w/T2w ratio, derived from conventional T1 and T2-weighted MRIs, is a particularly accessible proxy for myelin content [120] and shows lower values in WMHs (vs. NAWM) [57].

MWI offers greater specificity [121-123] than other qMRIs or DTI methods by quantifying the myelin water fraction (MWF) (Figure 1D) and the geometric mean T2 (GMT2), the latter indexing interstitial fluid content [71]. MWF is reduced within WMHs among stroke cohorts but not consistently in older adults without stroke [71], underscoring that demyelination cannot be inferred solely from DTI/qMRI, as other microstructural processes may predominate. In contrast, GMT2, indexing interstitial water, is elevated in WMHs [71], aligning with higher H₂O and FW. The combination of marked GMT2 increases with only modest or absent MWF reductions supports the view that fluid-driven alterations, rather than frank myelin loss, often dominate WMH pathology.

MTI provides a more direct assessment of macromolecular content than DWI, by quantifying magnetization exchange between protons bound to macromolecules (e.g., myelin) and those in the FW pool [36,66,124,125]. Its principal metrics include the magnetization transfer ratio (MTR) (Figure 1E) and the bound proton fraction (f_bound). Both MTR—reflecting macromolecular integrity [27,67,118]— and f_bound—myelin-related pool [118]—are reduced in WMHs compared with NAWM. Taken together, these modalities converge on evidence of myelin/macromolecular compromise in WMHs, although their specificity for demyelination per se remains limited and should be interpreted with caution.

BBB disruption is a key mechanism linking vascular pathology to WMH, mediated by endothelial tight-junction loss and elevated perivascular hydrostatic pressure that promote interstitial edema [84,91,126,127]. Dynamic contrast-enhanced MRI assesses BBB integrity by tracking the passage of intravenously administered gadolinium-based contrast agents [35,128,129]. In healthy tissue, the agent remains intravascular, whereas BBB disruption permits extravasation into the interstitium, producing time-dependent T1 signal changes.

Pharmacokinetic modeling of these curves yields quantitative indices of permeability and vascularity. Key parameters include the volume transfer constant (K_trans) and the permeability- surface area product (PS). Both K_trans and PS reflect contrast leakage. PS more directly indexes endothelial permeability, whereas K_trans is also influenced by perfusion flow rate; the plasma volume fraction (V_p) reflects intravascular plasma volume within tissue (Figure 1F). Increased K_trans or PS indicates endothelial tight-junction breakdown and BBB dysfunction, whereas elevated V_p suggests increased intravascular volume (e.g., vasodilation or angiogenesis) as compensation for reduced permeability or inflammatory hyperemia. Additional leakage metrics—the influx rate constant (K_i, leakage rate) and leakage volume (V_L, spatial extent of leakage)—further quantify BBB permeability. Across studies, K_trans [126], PS [84], K_i [91], and V_L [91] are consistently higher in WMH than in NAWM, indicating impaired barrier function. Notably, K_trans elevations are usually confined to PVWMH (vs. NAWM) and not consistently seen in DWMH, suggesting regionspecific mechanisms [126]. By contrast, V_p findings are heterogeneous—higher in cSVD cohorts [84], but lower values in cognitively normal/impaired older adults [126]—likely reflecting sensitivity of V_p to cSVD-related intravascular determinants (arteriolosclerosis, venous collagenosis, capillary remodeling/rarefaction, and vasomotor dysregulation) as well as methodological variability. An additional observation is the inverse relationship between K_trans or V_p and WMH volume [126]. Although counterintuitive given the association between global WMH burden and leakage [126], water-permeability imaging likewise shows a negative correlation between the BBB water exchange rate within WMH and WMH volume (r=-0.51) [127]. A plausible interpretation is that BBB leakage predominates early, whereas chronic lesions undergo tissue loss and reduced microvascular surface area, resulting in lower measured permeability and vascular volume. Analogous to multiple sclerosis, chronic WMHs may evolve toward atrophic [130], low perfusion states with diminished vascular metrics.

Chronic hypoperfusion also contributes to WMH formation [56,69,118,131-133] by inducing rarefaction of myelin sheaths [134-136] and glial activation [134,137-139] via excitotoxicity, oxidative stress, BBB dysfunction, and secondary inflammation [140,141]. Cerebral perfusion can be assessed with dynamic susceptibility contrast (DSC) MRI—which tracks contrast agents to quantify hemodynamics— and arterial spin labeling (ASL), which magnetically labels arterial water to quantify its delivery to brain tissue [37,142]. The principal metric is CBF, expressed as mL of blood delivered per 100 g of brain tissue per minute (Figure 1G). DSC also provides cerebral blood volume (CBV) and mean transit time (MTT) (Figure 1G)—the average duration of microvascular transit. Independent evidence from DSC-MRI and ASL consistently show reduced CBF [56,69,118,131-133] and prolonged MTT [112] within WMHs relative to NAWM, providing in vivo support for chronic hypoperfusion [143] as a major contributor to WMH development.

WMHs show metabolite alterations consistent with microstructural injury and impaired cellular energy metabolism [144-146]. MR spectroscopy (MRS) provides a non-invasive in vivo “chemical biopsy” of brain tissue by quantifying metabolite concentrations [38,147]. The most widely used technique, proton (1H) MRS, measures molecules such as N-acetylaspartate (NAA)—a marker of neuronal and axonal integrity [148,149]—choline (Cho), which reflects membrane turnover and increases with myelin breakdown or gliosis [148,150], and creatine (Cr), an index of cellular energy metabolism [148,151]. These metabolites are well-validated for reliable comparative quantification in both clinical and research contexts [152,153]. Within WMH, 1H-MRS typically shows reduced NAA (or NAA/Cr) [144,145] and reduced Cr [144] relative to NAWM, indicating diminished neuronal/axonal integrity (NAA) and impaired energy metabolism (Cr), respectively. A biphasic Cho/Cr pattern has been observed in the anterior horn region [145]: lower values in mild-moderate WMH but higher values in severe WMH compared with controls, suggesting early Cho elevation (reflecting membrane turnover or glial proliferation) followed by decline as irreversible injury accumulates and reparative capacity fails.

Reduced glucose uptake in white matter has also been linked to neuroglial dysfunction associated with WMH. 2-[¹⁸F]fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET) visualizes cerebral glucose metabolism [154,155], which normally parallels regional synaptic activity and declines with synaptic dysfunction or neurodegeneration. In white matter, axonal conduction and myelin maintenance impose substantial energetic demands on the oligodendrocyte-axon unit [156,157]; accordingly, WMH-related disruption of neuroglial support is expected to lower FDG uptake [68].

Multimodal MRI converges on a composite WMH pathobiology marked by tissue rarefaction and extracellular fluid expansion (FA↓, MD/RD↑, FW↑, T_c↑/T_w↓, GMT2↑, H₂O↑), variably accompanied by axonal injury (AD often ↑ with barrier loss, but ↓ in advanced collapse) and context-dependent demyelination (R1↓, T1w/T2w↓, MTR↓, V_MY↓; MWF↓ primarily in high-risk cohorts). Perivascular diffusion metrics and PVS enlargement indicate impaired glymphatic clearance (ALPS↓), suggesting failure of interstitial fluid drainage. Quantitative and magnetization-based measures converge on macromolecular compromise, while vascular signatures reveal BBB dysfunction (K_trans/PS/K_i/V_L↑), heterogeneous V_p behavior across cohorts, and chronic hypoperfusion (CBF↓, MTT↑; D* tending ↓, f often ↑ via compensatory vasodilation). Metabolic imaging demonstrates impaired neuronal/axonal integrity and energy metabolism (NAA↓, Cr↓, FDG-PET regional standardized uptake value [rSUV]↓) with stage-dependent choline shifts reflecting early gliosis followed by exhaustion of reparative capacity. Collectively, these modalities depict WMH as a multifactorial process integrating microstructural degradation, vascular and glymphatic failure, and metabolic insufficiency.

Relative to NAWM, both PVWMH and DWMH exhibit lower FA and higher MD, AD, and RD [27,67-73], consistent with reduced fiber coherence and extracellular space expansion. Across multiple studies, FA is consistently lower in PVWMH than in DWMH [54,72,165]. One study that separated JVWMH (<3 mm from the ventricles) reported a gradient of FA reduction with ventricular proximity (DWMH > PVWMH > JVWMH) [54]. T_c increases and T_w decreases closer to the ventricle (from DWMH to JVWMH) [55], indicating a shift toward a CSF-like composition, consistent with axon/myelin loss and interstitial fluid expansion. Another study [165] found particularly low FA in frontal PVWMH compared with occipital or parietal PVWMH, underscoring heterogeneity within PV regions. By contrast, MD, AD, and RD were highest in JVWMH, lowest in PVWMH, and intermediate in DWMH (JVWMH > DWMH > PVWMH), suggesting a potentially nonlinear ventricle-to-cortex gradient. A recent study [84] also reported higher FW (and MD) in PVWMH than in DWMH, supporting greater interstitial fluid accumulation and matrix rarefaction near the ventricles. Taken together, PVWMHs generally show more severe microstructural rarefaction and higher water content than DWMH, although PVWMHs located farther from the ventricles (>3 mm) may exhibit smaller changes—sometimes even less than those seen in DWMH.

IVIM-derived D is generally higher in WMH than in NAWM [87,90,91]. Notably, one study [90] reported greater D in PVWMH than in DWMH, consistent with higher interstitial water content attributed to increased BBB permeability and plasma leakage in PVWMH [166].

Finally, the DTI-ALPS index—an estimate of diffusion along perivascular pathways reflecting glymphatic clearance—was independently associated with DWMH volume but not PVWMH volume [167], consistent with histopathological evidence [21,159,162-164] of more prominent enlarged PVS in deep white matter than in periventricular regions. By contrast, PVWMH volume was related to ALPS only indirectly via deep medullary vein remodeling: impaired perivascular clearance (lower ALPS) may promote venous injury and remodeling on a background of frequent periventricular venous collagenosis [160,161], thereby compromising drainage and fostering interstitial edema and lesion expansion. DWMHs are frequently (70%-90%) located adjacent to enlarged deep white matter PVS (dwPVS) [105,106], and their volume scales with enlarged dwPVS burden [105]. In contrast, PVWMHs show limited proximity to PVS (including enlarged basal ganglia PVS [bgPVS]), indicating that PVS-related pathology is more characteristic of DWMH than PVWMH [105]. Nonetheless, multiple cohorts report that enlarged bgPVS severity correlates with overall WMH burden [168-171], an association that is not location-specific and may reflect a more global microvascular or clearance dysfunction. Notably, FW within DWMH mediated the relationship between enlarged dwPVS and DWMH volume [105], further supporting a mechanistic link via impaired interstitial fluid drainage.

Across myelin-related qMRI metrics, both PVWMH and DWMH show lower R1 (1/T1), R2* (1/T2*), V_MY, and T1w/T2w, together with higher H₂O, compared with NAWM [27,57,118,119]. Notably, PVWMH exhibits larger deviations than DWMH—lower R1 and R2* and higher H₂O—suggesting more pronounced myelin/macromolecular alterations toward the ventricles along the ventricle-to-cortex gradient [118]. Consistent with this, V_MY is lower in PVWMHs than in DWMHs [119], supporting greater myelin/macromolecular compromise in periventricular regions. MTI-based indices show a similar pattern, with both MTR [118,172] (macromolecular integrity) and f_bound [118] (myelin-related pool) reduced in PVWMHs relative to DWMHs. However, direct MWI comparisons between PVWMHs and DWMHs remain scarce; accordingly, these qMRI and MTI differences should not be over-interpreted as specific evidence of demyelination.

Notably, declines in myelin/macromolecular indices (lower R1 or T1w/T2w) predict cross-sectional WMH volume (or Fazekas grade) within DWMH or JCWMH, but not within PVWMH [57,118]. By contrast, PVWMH volume (or Fazekas grade) shows a stronger association with lower R2* [57,118]—a metric influenced by paramagnetic iron concentrated in oligodendrocytes [173,174]. This dissociation suggests that demyelination and macromolecular degradation may be more central to DWMH/JCWMH progression, whereas PVWMH progression may be more tightly linked to iron-related oligodendrocyte vulnerability. Accordingly, lower qMRI values in PVWMH relative to DWMH should not be taken as evidence of greater myelin/macromolecular damage per se, but may instead reflect ventricle-adjacent CSF or iron effects and local vascular/ependymal pathology.

Across studies, K_trans, PS, K_i, and V_L are consistently elevated in WMH compared with NAWM, indicating increased BBB leakage [84,91,126]. K_trans is also higher in PVWMH than in DWMH, suggesting a more prominent contribution of BBB dysfunction to PVWMH pathobiology [126]. By contrast, V_p shows divergent results across cohorts—higher [84] in PVWMH among patients with cSVD but lower [126] in PVWMH among cognitively normal and impaired older adults—supporting the interpretation that V_p is sensitive to intravascular determinants (such as microvascular volume, arteriolosclerosis, venous remodeling) beyond BBB leakage, potentially accentuated near the ventricular CSF interface [35,175].

Multiple studies report lower regional CBF in PVWMH than in DWMH [56,69,176], with the lowest values in JVWMH [56]. However, these absolute CBF differences across JVWMH, PVWMH, and DWMH should be interpreted with caution. MRI-derived CBF is vulnerable to biases near the ventricles—prolonged arterial transit time and CSF partial-volume effects—that can artifactually depress periventricular estimates independent of true hypoperfusion [133,177]. Additionally, partial-volume contamination from gray matter—given the large gray matter-white matter perfusion contrast—can bias white matter CBF estimates, particularly in deep WMHs where voxels border deep gray nuclei (e.g., thalamus, basal ganglia), leading to artificially elevated WM perfusion values [178,179]. Consistent with this, one study reported a distance-to-ventricle effect, with progressively lower CBF in both WMH and perilesional NAWM closer to the ventricles [56]. Even so, lower CBF is also linked to perilesional NAWM abnormalities (penumbra) [78,176,180,181], new WMH [56,69], and WMH lesion growth [69] in both PVWMH and DWMH, challenging the notion that chronic hypoperfusion is primarily a DWMH pathology. Plausible explanations include (1) sampling and interpretive limitations of prior histopathology and (2) tight coupling of CBF reductions to periventricular processes (interstitial fluid accumulation, BBB leakage), producing confounding. Future work should combine histopathologically anchored designs with confounder-controlled MRI perfusion methods—such as multi-post-labeling delay (arterial transit time-corrected) ASL—and integrate BBB and water-content indices to isolate hypoperfusion-specific effects.

Across modalities, PVWMH and DWMH share a core signature of microstructural injury (FA↓, MD/RD↑) but the primary type of damage differs. PVWMHs show greater extracellular fluid shifts (FW↑; diffusion composition T_c↑/T_w↓), more BBB leakage (K_trans↑), and larger deviations in myelin/macromolecular indices (R1/R2*/V_MY/MTR↓). This pattern is consistent with a periventricular interstitial-edema phenotype with BBB leakage and venous remodeling. By contrast, DWMHs are closely associated with PVS enlargement and lower ALPS, indicating impaired glymphatic drainage, while deep-perforator ischemia also contributes. Together, multimodal MRI indicates that PVWMHs are dominated by edema/clearance failure and BBB dysfunction, whereas DWMHs represent a PVS-associated glymphatic phenotype with a contributory role of deep perforator ischemia.

Perilesional NAWM shows a distance-dependent, inverted-Ushaped FA profile: FA rises from distant NAWM toward ~4 mm from the WMH margin, then declines from ~4 mm to the lesion edge [27,28,70,71,78,180,181]. Accordingly, perilesional NAWM may show either higher [27,28,70,78,181] or lower [26,67,71,180] FA than distant NAWM depending on the sampling radius and study design. When PVWMHs and DWMHs are analyzed separately, the inverted-U profile is more pronounced around PVWMH [78,180,181], likely reflecting location effects—PVWMHs abut highly coherent commissural and association fibers (e.g., corpus callosum) where diffusion anisotropy is intrinsically high [27]. Other diffusion parameters (MD/AD/RD) are generally elevated in perilesional NAWM compared with distant NAWM [27,28,67,71,78,84,180,181]. Penumbra size defined by MD (or M_K) [28,71,78,84,180,181] ranges 2-12 mm, with most estimates 4-7 mm; in the single study [78] quantifying AD (or A_K) and RD (or R_K) gradients, the AD-defined penumbra extended 5-14 mm and the RD-defined penumbra 6-11 mm. Across studies [78,84,180,181], PVWMH and DWMH differ in penumbra extent, although the direction varies across datasets and metrics, reflecting heterogeneity in lesion topography and penumbra definitions. Larger samples, finer spatial classification, and harmonized criteria are needed to resolve these discrepancies. Consistent with diffusion metrics, FW is higher in perilesional NAWM compared to distant NAWM [70,84], with penumbra size estimated at ~4-8 mm and similar between PVWMHs and DWMHs. A decreasing IVIM f toward the WMH edge, together with rising FW, supports a pattern of predominantly extravascular fluid accumulation rather than increased intravascular volume [84]. In one longitudinal study [91], D increased over 2 years with a distance-dependent gradient—greater in perilesional NAWM than in distant NAWM—indicating faster microstructural damage in the perilesional zone, consistent with prior DTI findings.

In NAWM immediately adjacent to PVWMH (<2 mm), R1 (1/T1) [27,28] is higher but MTR [27,67] is lower than in distant NAWM, with both measures lowest within WMH—yielding an inverted-U profile similar to that observed for DTI-FA. Although direct PVWMH-DWMH comparisons of R1/MTR are limited, this pattern likely reflects a location effect, given the predominance of PV-adjacent lesions. One study [67] also reported greater MTR reduction in frontal NAWM adjacent to PVWMH compared with parietal-occipital NAWM, highlighting spatial heterogeneity within periventricular regions and reinforcing the role of location effects. Notably, MWI demonstrated higher GMT2 (interstitial water) in perilesional NAWM (<6 mm) than in distant NAWM, whereas MWF did not differ across 2-10-mm shells, suggesting that fluid-driven changes, rather than overt demyelination, may predominate in the early penumbra [71].

Perilesional NAWM exhibits a distance-dependent rise in BBB leakage, increasing from distant NAWM toward the WMH margin and peaking within the lesion [28]. In one study, PS and V_p showed little variation across perilesional NAWM [84], whereas other studies reported progressive increases in K_i [91] and V_L [91,182] with proximity to the lesion, reaching their highest values within WMH. Notably, baseline BBB leakage (K_i and V_L) correlated with 2-year increases in D in the perilesional zone [91], supporting the hypothesis that early BBB impairment contributes to subsequent white matter degeneration. The minimal gradients in PS and V_p, contrasted with significant K_i and V_L changes, are most plausibly explained by differential model sensitivities for BBB permeability estimation (Tofts vs. Patlak) [183]. Further studies are needed to clarify how these metrics diverge in practice and which offers superior prognostic value for BBB dysfunction in WMH.

Multiple studies consistently reported lower regional CBF in perilesional NAWM than in distant NAWM, with the lowest CBF values observed within WMH [56,78,176,180-182]. The CBF-defined penumbra typically spans 7-14 mm and appears broadly similar for PVWMH and DWMH. In studies that derived DTI-defined (structural) penumbras in parallel [78,180,181], the CBF-defined penumbra was generally broader; in one study [78], however, this divergence was observed only for DWMH, a finding that warrants more precise replication.

Across modalities, perilesional NAWM functions as an intermediate injury zone with distance-dependent gradients that normalize with increasing separation from the WMH core. Microstructural rarefaction and extracellular fluid expansion intensify near lesions (MD/AD/RD↑, FW↑), while FA (and R1) follows an inverted-U profile peaking around 4 mm from the margin. In PV-adjacent NAWM, myelin/macromolecular indices (e.g., MTR) mirror this pattern, whereas elevated GMT2 without clear MWF reduction indicates that fluid-driven alterations, rather than overt demyelination, dominate early injury. Vascularly, BBB leakage (K_i, V_L) increases toward the lesion edge, and baseline leakage predicts subsequent D increases (ΔD), linking early barrier dysfunction to progressive microstructural decline. Perfusion deficits (CBF↓) extend beyond the structural (diffusion) penumbra (~7- 14 mm vs. ~2-6 mm), suggesting that hemodynamic compromise outpaces tissue degeneration. Collectively, these findings support the view that perilesional NAWM forms a dynamic continuum of injury—a pathophysiologically informative zone at risk that underscores the importance of harmonized definitions, spatially resolved analyses, and integrative multimodal metrics to improve prognostic assessment.

Multiple studies consistently reported lower baseline FA [29,31,68,69,184,185] and higher baseline MD [31,68,69,185] in regions that subsequently develop into new WMHs compared with persistent NAWM. One study [69] observed higher baseline AD and RD, together with lower FA and higher MD, in both PVWMH and DWMH. Notably, baseline RD was positively associated with subsequent WMH growth across both lesion types.

Extending cross-sectional observations of the spatial proximity between DWMHs and enlarged dwPVS [105,106], a longitudinal study found that ~70% of new DWMHs emerged around enlarged dwPVS at follow-up [106]. This pattern supports the interpretation that a major component of DWMH pathobiology reflects impaired interstitial fluid drainage along perivascular pathways.

Baseline CBF is lower in regions that develop into new WMHs than in persistent NAWM [69,132]. One study found this reduction confined to JVWMH and PVWMH, but not DWMH [56], while another [69] reported that baseline CBF predicted PVWMH growth but not DWMH growth—consistent with a location effect.

Blood oxygenation level-dependent (BOLD) functional MRI provides an indirect measure of cerebral hemodynamics by detecting T2*-weighted signal changes driven by deoxyhemoglobin [186,187]. A central application is the assessment of cerebrovascular reactivity (CVR)—the BOLD response to a standardized vasoactive stimulus [188-190]. CVR can be measured from the BOLD changes by simpler breath-hold tasks [188,189] or using controlled gas paradigms that elevate end-tidal CO2 while maintaining end-tidal O2 [190], thereby inducing hypercapnia. As a vascular reserve biomarker, CVR—including its steady-state component (ss-CVR), representing the magnitude of the vasodilatory response, and its dynamic component (tau), reflecting the speed of vascular adjustment—complements static perfusion measures by probing the capacity of the microvasculature to dilate [191]. These BOLD-based metrics distinguished future WMHs from persistent NAWM [185,191], suggesting that impaired vasodilatory reserve, beyond low CBF, contributes to WMH progression.

One MRS study demonstrated longitudinal reduction in NAA and Cr—but not glutamate/glutamine, Cho, or myo-inositol—within WMHs, suggesting that neuronal/axonal integrity (NAA) and energy metabolism (Cr) are compromised in persistent lesions and may contribute to progression risk [146]. Complementing MRS, FDGPET demonstrates lower baseline rSUV in new WMHs than in persistent NAWM [68], indicating that metabolic impairment accompanies structural WM injury.

Across longitudinal studies, new WMHs show a convergent baseline signature of microstructural compromise (FA↓, MD/AD/RD↑) most pronounced in regions that subsequently enlarge—linking pre-existing tissue rarefaction to lesion formation. Baseline RD further predicts future growth, supporting its utility as a microstructural risk marker. Hemodynamically, new lesions originate from regions with reduced baseline CBF, while impaired vasodilatory reserve on BOLD-CVR (including ssCVR and tau) prospectively distinguishes tissue destined to convert from tissue that remains NAWM, indicating that vascular reactivity failure, beyond low flow alone, contributes to progression. Topographically, new DWMHs cluster around enlarged dwPVS, consistent with impaired interstitial fluid drainage as a location-specific mechanism. Metabolically, longitudinal reductions in NAA and Cr within persistent lesions indicate ongoing neuronal/axonal injury and energy dysfunction, while lower baseline FDG-PET rSUV in new WMHs supports metabolic disturbance as an additional driver of progression. Collectively, these observations underscore that WMH progression is a complex, multicomponent process shaped by interactions among microstructural vulnerability, vascular dysregulation, impaired perivascular drainage, and metabolic stress.

The ventricle-to-cortex gradient emerges as a fundamental organizing principle for WMH pathobiology. Evidence indicates that PVWMHs and DWMHs represent mechanistically distinct entities rather than variants of a single process.

PVWMHs exhibit a characteristic profile of periventricular interstitial edema, marked by prominent BBB dysfunction (elevated K_trans, PS, K_i, and V_L) [84,91,126], extensive extracellular fluid accumulation (increased FW and T_c) [55,84]. This pattern aligns with histopathological evidence of ependymal disruption, subependymal gliosis, and periventricular venous collagenosis [17,19,21,22,158-161]. Moreover, the association between PVWMH progression and R2* reductions [57,118]—reflecting iron-related oligodendrocyte vulnerability—further distinguishes periventricular pathobiology from deep white matter changes.

In contrast, the DWMH phenotype is more robustly characterized by glymphatic dysfunction. A direct association between the DTI-ALPS index and DWMH volume [167], together with the frequent colocalization of DWMHs with enlarged dwPVS [105,106], implicates clearance failure as a central mechanism. FW further mediates the relationship between dwPVS burden and DWMH volume [105], supporting a pathway in which impaired interstitial—perivascular drainage promotes extracellular fluid accumulation and lesion growth. Myelin and macromolecular integrity indices (R1 and T1w/T2w) also predict DWMH and JCWMH progression but not PVWMH volume [57,118], suggesting that demyelination and macromolecular degradation are more prominent contributors to deep-lesion expansion.

Finally, the observation that MRI-derived CBF reductions are not specific to DWMH challenges the prevailing view of selective deep-perforator hypoperfusion and calls for re-examining vascular mechanisms across the ventricle-to-cortex continuum [56,69,78,176,180,181]. However, despite the limited specificity of CBF measurements alone, epidemiological and clinical studies have shown that DWMH correlate more strongly with cardiovascular mortality [192], recurrent stroke (with larger risk ratios than PVWMH),193,194 and vascular dementia [195,196], whereas PVWMH exhibit closer associations with noncardiovascular mortality [192] and Alzheimer’s disease [195,196]. These divergent patterns suggest that vascular pathology beyond BBB leakage—particularly ischemia-related demyelination and impaired clearance—may be more tightly linked to DWMH. Similarly, our recent work [15] in a consecutive stroke cohort (in which PVWMH predominated) [6] found that total WMH burden was more closely related to nonvascular mortality and hemorrhagic stroke recurrence. Thus, although ischemia is not exclusive to DWMH, these lesions may display greater susceptibility under comparable ischemic stress, a plausible hypothesis that merits prospective validation.

Longitudinal MRI signatures delineate a temporal cascade of pathobiological events driving WMH formation and progression. Pre-lesional tissue destined to evolve into WMH exhibits a characteristic baseline profile: microstructural compromise (lower FA and higher MD, AD, and RD) [27,28,67,70,71,78,84,180,181], reduced perfusion (lower CBF) [56,78,176,180-182], and impaired vascular reserve (abnormal CVR, ssCVR, and tau) [185,191]. Notably, baseline RD predicts subsequent WMH growth [69], establishing it as a mechanistic biomarker that links early myelin/axonal injury to later lesion expansion.

The perilesional penumbra constitutes a critical transition zone where pathobiological processes evolve. The observation that the hemodynamic penumbra (7-14 mm) is broader than the structural penumbra (2-6 mm) [78,180,181] suggests that perfusion deficits precede overt tissue damage. The correlation between baseline BBB leakage (K_i, V_L) and subsequent increase in D in perilesional NAWM [91] provides direct evidence for a vascular-driven progression toward structural injury. However, future studies investigating the structural penumbra should incorporate myelin-sensitive markers (e.g., R1, MWF, MTI) alongside conventional water diffusion metrics (e.g., FA, AD, MD, RD). While diffusion metrics are sensitive to microstructural disruption, they lack specificity and can be altered by ischemia-related changes in intravascular and plasma volumes even in the absence of substantial demyelination [28,69,78,84,180,181]. Within this framework, the inverted-U profile of FA and myelin indices (R1 and MTR) in perilesional NAWM likely reflects competing processes [27,28,70]: compensatory tissue reorganization at intermediate distances from the lesion and progressive degeneration near the lesion edge.

Identifying a patient’s predominant spatial phenotype has clinical relevance, as it reflects distinct underlying mechanisms that may demand different monitoring priorities and interventions.

Longitudinal MRI signatures provide prognostic information by delineating the temporal cascade of WMH injury, thereby defining a window for targeted prevention and surveillance. The key is identifying tissue at risk, most evident within the perilesional penumbra. Notably, the hemodynamic penumbra—defined by reduced CBF—is typically broader (~7-14 mm) than the structural penumbra defined by diffusion changes (~2-6 mm) [78,180,181], suggesting that perfusion deficits precede overt tissue injury. This sequence from vascular dysfunction to subsequent structural damage is further supported by evidence that BBB leakage in the penumbra predicts subsequent microstructural decline [91]. However, this cascade does not imply that all axonal or myelin loss in WMH is ischemic in origin. Non-ischemic demyelination can occur independently of vascular insufficiency, as exemplified by autoimmune demyelinating diseases such as multiple sclerosis [121,124,125,215,216]. Non-ischemic demyelination within WMH may present with disproportionate reductions in myelin-sensitive measures despite relatively preserved perfusion, suggesting primary myelin loss rather than secondary ischemic injury. Accordingly, integrating quantitative myelin indices with perfusion metrics could distinguish whether demyelination in a given WMH arises from ischemic injury, primary demyelinating processes, or mixed pathology—thereby enabling a more comprehensive understanding of WMH progression mechanisms.

These vascular and microstructural signatures may assist clinicians in identifying patients on an accelerated trajectory of WMH accumulation. For individuals at high risk based on longitudinal MRI markers, early treatment should target initial pathobiological processes—impaired perfusion and BBB dysfunction—through intervantions such as intensive blood-pressure control [197-199] and pharmacologic enhancement of cerebrovascular function [212,213] before substantial tissue loss ensues. Because total WMH volume evolves slowly and is an insensitive short-term marker, follow-up should emphasize more responsive surrogates, tailored to clinical feasibility. Non-contrast ASL perfusion can detect hemodynamic changes preceding WMH growth [56,78,176,180-181]. and remains valuable for ischemic penumbra assessment in stroke [209-211]. When advanced imaging is unavailable, serial FLAIR at 1-2-year intervals, assessed by visual grading or volumetry, can still differentiate rapid from slow progressors and guide preventive intensity [217].

Stroke patients with advanced WMH accumulation face higher risks of poor functional outcomes, stroke recurrence, and mortality. Our nationwide multicenter studies demonstrated that higher baseline WMH burden independently worsens outcomes after ischemic stroke [14,15,218], underscoring its prognostic value for secondary prevention. The detrimental impact of WMH volume is particularly pronounced in mild stroke, where WMH burden strongly predicts 3-month functional outcomes, whereas this association weakens in moderate to severe stroke [218]. Thus, active monitoring and management of WMH progression may be more important in mild stroke. In addition, although WMH distribution is typically symmetric between hemispheres, we observed that asymmetric WMHs were associated with both old silent and acute lacunar infarcts ipsilateral to the hemisphere with greater WMH burden, supporting the inclusion of WMH asymmetry in clinical risk assessment [53].