Diffusion kurtosis imaging detects cortical microstructural alterations in amyloid-positive MCI patients

Background: Alzheimer's disease (AD) is characterized by early accumulation of amyloid-β (Aβ) plaques and tau pathology which precede overt neurodegeneration and cognitive decline. Detecting microstructural brain changes associated with Aβ deposition before the onset of atrophy is critical for early diagnosis and intervention.

Objective: This study investigates whether diffusion kurtosis imaging (DKI) can detect early microstructural alterations in cortical and subcortical gray matter (GM) associated with Aβ pathology in individuals with mild cognitive impairment (MCI).

Methods: Using DKI-derived metrics—mean kurtosis (MK) and mean diffusivity (MD) – we assessed cortical and subcortical microstructure in 67 participants (23 cognitively normal [CN], 44 MCI, including 29 Aβ-positive). Aβ burden was quantified using ¹¹C-PiB PET imaging. Cortical atrophy, hippocampal volume, and white matter hyperintensities (WMH) were also evaluated.

Results: Aβ-positive MCI patients exhibited significantly elevated cortical MK, particularly in the left lateral temporal lobe and right precuneus, compared to both CN and Aβ-negative MCI groups. MK positively correlated with Aβ burden in parietal and temporal cortices, even in the absence of cortical atrophy. In contrast, MD showed weaker and less consistent associations with Aβ and was more strongly influenced by age. No significant subcortical MK or MD differences were observed.

Conclusion: Elevated MK in Aβ-positive MCI patients suggests that DKI can detect early microstructural changes associated with the presence of amyloid pathology before the onset of cortical atrophy. MK may serve as a promising non-invasive biomarker for identifying prodromal AD and monitoring disease progression.

1 Introduction

The neuropathological features of Alzheimer's disease (AD) include plaques of extracellular cerebral amyloid β (Aβ) protein and intracellular neurofibrillary tangles of hyperphosphorylated tau protein. The accumulation of Aβ is thought to antedate cognitive decline and detectable cerebral atrophy by many years (Jack et al., 2013). Positron emission tomography (PET) offers a means of detecting cerebral fibrillar Aβ and tau tangles several years before cognitive symptoms arise, while magnetic resonance imaging (MRI) is sensitive to brain atrophy in the form of hippocampal volume reduction and cortical thinning (Frisoni et al., 2010). We currently lack tools to non-invasively detect and characterize the cellular injury caused by cerebral protein deposits in the years before overt brain atrophy and clinical AD develops. Such tools are crucial to improve our understanding of AD's preclinical phases and to offer early, preventative therapies to relevant patient cohorts.

Diffusion-weighted MRI (DWI) is sensitive to the micrometer-scale Brownian motions of water molecules and, thereby, to tissue microstructure as these water molecules collide and interact with cell membranes and organelles (Jensen and Helpern, 2010; Jespersen et al., 2010). Indeed, DWI suggests that changes in cerebral microstructure precedes atrophy in subjects with mild cognitive impairment (MCI; Sexton et al., 2011; Struyfs et al., 2015; Weston et al., 2015; Taoka et al., 2016; Gong et al., 2017; Lee et al., 2017; Arab et al., 2018; Gyebnar et al., 2018). By measuring diffusion across multiple directions, it is possible to deduct information about the underlying structural directionality. This technique, dubbed diffusion tensor imaging (DTI), is particularly useful for assessing integrity of white matter (WM) fiber tracts. Nevertheless, in gray matter (GM) where neurotoxic Aβ oligomers and p-tau proteins accumulate in AD, the overall tissue architecture is far more complex with dendrites and axons following less discernible structural patterns. It is feasible to characterize the GM microstructure in terms of the mean diffusivity (MD) under the assumption that water molecules diffuse freely in the tissue. In theory, any microstructure that hinders water-diffusion reduces MD. Conversely, increased MD indicates a less restricted diffusion process potentially associated with neurodegenerative changes such as loss of cells, dendrites, and axons or increased free extracellular water content (Basser and Pierpaoli, 1996; Kale et al., 2006; Weston et al., 2015).

Nonetheless, describing the complex microstructural organization of brain tissue only by a Gaussian model, as in mean diffusivity (MD) from DWI and DTI, remains a considerable oversimplification. Diffusion kurtosis imaging (DKI) addresses this limitation by quantifying the extent to which water diffusion deviates from a Gaussian distribution. The key parameter, mean kurtosis (MK), reflects the average degree of non-Gaussian diffusion and thus serves as an indicator of microstructural complexity. Compared with DTI, DKI requires acquisition at multiple and higher b-values (Jensen and Helpern, 2010). MK may potentially provide additional sensitivity toward early erratic loss of cells, dendrites, and axons in AD, which might not be detected with plain MD (Jensen and Helpern, 2010). Raised MK has been observed in regions with high concentration of Aβ plaques in the transgenic APP/PS1 mouse model (Vanhoutte et al., 2013; Praet et al., 2018). This suggests a role for MK as an early, non-invasive, and relatively cost-effective in-vivo biomarker of cellular injury in brain regions associated with neurotoxic amyloid pathology.

In this study, we measured cortical and subcortical gray matter MK and MD using DKI to examine their microstructural changes and the association with Aβ deposition in subjects with MCI.

2 Materials and methods

The Regional Ethics Committee for Biomedical Research in the Central Denmark Region approved the study [1-10-72-116-13] in agreement with the declaration of Helsinki, and participants signed an informed written consent ahead of enrolment.

2.1 Participants

Subjects with MCI who fulfilled the Petersen criteria (Petersen, 2004) and normal controls (CN) with no cognitive complaints were included in this study. All participants were recruited from national memory clinics and through newspaper advertisements. Please refer to Parbo et al. (2017) for a detailed description of recruitment and inclusion- and exclusion-criteria. We classified participants as Aβ-positive (Aβ+) or Aβ-negative (Aβ-), based on visual reads by experts and a dichotomizing threshold region of interest (ROI) mean ¹¹C Pittsburgh Compound B standard uptake value ratio (PiB SUVr) set at 1.5, which naturally separated the bimodal distribution of composite cortical PiB SUVrs into low and high subgroups. In addition, all participants were assessed with MRI and a series of standardized neuropsychological tests recommended for identifying dementia, early cognitive decline and depression, as previously described (Parbo et al., 2017). We present summary scores for Clinical Dementia Rating (CDR), Mini-Mental State Examination (MMSE), and Geriatric Depression Scale (GDS-15). A GDS-15 score of 6 or more was considered an exclusion criterion.

2.2 Brain imaging and preprocessing

MRI was acquired on a Siemens Magnetom 3T Skyra system (Siemens Healthcare, Erlangen, Germany) using a 32-channel head coil. DKI was acquired using a b-value scheme optimized for clinical settings, offering both speed and robustness (Hansen et al., 2013, 2014), including one b = 0 image, three b = 1000 s/mm² and nine b = 2500 s/mm² with 2.3 mm isotropic voxels, FOV = 220 × 220 mm² in 38 slices, TR = 12.4 s, TE = 0.107 s and TI = 2.1 s. The duration of the sequence is less than two minutes reducing the risk of movement artifacts. In addition, anatomical T₁-weighted magnetization-prepared two rapid gradient echo (MP2RAGE; Marques et al., 2010) and T₂-weighted fluid attenuated inversion recovery (T₂-FLAIR) MRI were acquired. See Supplementary Material for additional information on MRI sequence settings.

Experienced radiologists screened the anatomical images to ensure exclusion of subjects with structural abnormalities, including tumors, large infarcts, hydrocephalus, and other major structural brain pathology that could significantly impact cognition. Additionally, the degree of chronic small vessel ischemia as indicated by the extent of visible WM hyperintensities (WMH) on T₂-FLAIR images was graded using the Fazekas scale and quantified by WMH volume segmented using an automatic algorithm (Schmidt et al., 2012).

MD and MK were calculated from DKI with the method described by Hansen and colleagues (Hansen et al., 2013, 2014), using denoised (Veraart et al., 2016) images corrected for Gibbs ringing artifacts (Kellner et al., 2016). To localize cortical and subcortical MD and MK, the DKI were aligned to the MP2RAGE using the b = 0 image of the DKI sequence (Avants et al., 2011).

Cerebral amyloid accumulation was characterized visually and as PiB SUVrs, calculated from PiB PET images acquired on a High-Resolution Research Tomograph (ECAT HRRT; CTI/Siemens, Knoxville, TN, USA). As previously described (Parbo et al., 2017), the cerebellar GM mean ROI was used as a non-specific reference. PET images were aligned with corresponding anatomical T₁-weighted MRIs to localize cortical PiB SUVrs (Collins et al., 1994).

2.3 Atrophy, cortical measurements and white matter hyperintensity volume

First, we measured hippocampal volume (Coupe et al., 2011) and cortical thickness to detect atrophy. The latter was defined as the perpendicular distance between the inner and outer cortical borders delineated on MP2RAGE (Eskildsen and Ostergaard, 2006; Aubert-Broche et al., 2013). Then, we extracted cortical values of MK, MD and PiB SUVr at the middle point between these inner and outer cortical borders, using methods described previously (Eskildsen et al., 2017; Nielsen et al., 2017). By sampling along the central cortical layers, partial volume effects were minimized.

WMH volume was measured as the volume of T₂w-FLAIR WMHs, segmented using a histogram-based method (Smart et al., 2011) followed by region growing and manual inspection and corrections of segmentations. Subject-specific intracranial volume was used for normalization of all volumetric measurements (Eskildsen et al., 2012).

2.4 Subcortical diffusion

Cerebral regions involved in the frontal subcortical circuits were automatically segmented, using an atlas (Neuromorphometrics, Inc., Somerville, Massachusetts) located in MNI space. Subcortical regions included bilateral segmentations of the hippocampus, amygdala, thalamus, nucleus accumbens, nucleus caudatus, putamen, and globus pallidus. The atlas was moved into native DKI space for extraction of MK and MD mean ROI values (SPM 12 version 7219, The Wellcome Trust Center for Institute of Neurology, University College London).

2.5 Statistical analysis

Statistical computations were performed with R version 4.3 (R Foundation for Statistical Computing, Vienna, Austria). Group differences in cortical and subcortical ROI mean values of MK, MD, and PiB SUVr were assessed using linear regression. Likewise, linear regression was applied to assess correlations between cortical PiB SUVr measurements and cortical MK and MD measurements, respectively. In both cases, we included age as a covariate since aging is known to affect diffusional parameters (Falangola et al., 2008; Gong et al., 2014; Taha et al., 2022). In separate models we additionally adjusted for sex and MMSE scores as some studies have found these to influence diffusion, though mainly in WM (Lawrence et al., 2021; Matsumoto et al., 2025). For the cortical analyses we visualized t-values for p < 0.05 (one-tailed) and outline clusters surviving family wise error correction (FWE) for multiple comparisons (Worsley et al., 1996) at α = 0.001.

Demographic differences across the three groups of CN subjects, Aβ+ MCI patients and Aβ- MCI patients, respectively, were assessed in a two-step procedure. First, we tested for any differences across all three groups. If any group differences were detected, we compared individual subgroups. For continuous variables following a normal distribution, we assessed differences across all groups using ANOVA, while Tukey's test for post hoc multiple pair-wise comparisons interrogated differences between individual groups. Otherwise, we applied the Kruskal–Wallis test across all three groups and post hoc Wilcoxon rank-sum test with Bonferroni correction for the individual pairwise group comparisons. Categorical variables were evaluated across all three groups with the extended Fischer's exact test, including post hoc pairwise comparisons and Bonferroni correction for comparison of individual groups. We considered p < 0.05 (two-tailed) statistically significant.

3 Results

Twenty-three CN subjects and 44 MCI patients (29 Aβ+) were included in the study. Their demographic characteristics, neuropsychological test results, WMH volume and hippocampal volume are summarized in Table 1. On average, Aβ+ MCIs were older than Aβ- MCIs, and they scored lower than both Aβ- MCIs and CNs on cognitive tests. Aβ- MCIs performed worse than CNs on cognitive tests with a significantly raised mean CDR sum-of-boxes score. Antihypertensive drug use was more common among Aβ+ MCIs, but all groups showed similar Fazekas scores and WMH volumes, indicating similar vascular lesion load across subjects. We only observed a trend toward reduced hippocampal volume in MCIs compared to CNs. No obvious cortical thinning was detected in any of the MCI groups compared to the CN group (Supplementary Figure S1).

Table 1

Table 1. Demographics, neuropsychological test results, hippocampal volume and WMH load according to subgroup.

3.1 Mean kurtosis

Cortical MK was significantly elevated in the Aβ+ MCI patients compared to the CNs, with clusters surviving FWE correction in parts of the left lateral temporal lobe and in the caudal part of the right precuneus close to the parietooccipital sulcus (Figure 1). The relative changes extended even further across the lateral parts of the left temporal lobe in the Aβ+ MCI patients when compared to the Aβ- MCI patients, indicating an association between Aβ pathology and MK (Figure 1). In further support of this association, MK was trending lower in the Aβ- MCI group compared with the group of CNs which contained some subjects with undetermined Aβ status (Figure 1). Extending the linear models with sex and MMSE only slightly modified the statistical maps without changing the interpretation (Supplementary Figure S3).

Figure 1

Figure 1. Statistical t-maps of significant (p < 0.05) differences in cortical mean kurtosis between patients with mild cognitive impairment (MCI) and cognitively normal controls and between subgroups of MCI patients. Negative t-values indicate decreases (blue nuances), while positive t-values indicate increases (red nuances) in patients. The white outlining marks clusters surviving family-wise error correction for multiple comparisons (p < 0.001). Age-adjusted region of interest mean values from clusters with the highest mean t-value, as indicated with the black arrows, are shown in the lower panel, including mean (black dot) and standard error of the mean (whiskers). Patients were sub-grouped as amyloid positive (Aβ+) or amyloid negative (Aβ-) based on their ¹¹C-PiB (Pittsburgh Compound B) status, defined by an atlas based composite regional ¹¹C-PiB SUVr (standard uptake value ratio) >1.5 and ≤ 1.5, respectively. Statistics were adjusted for age using linear regression.

In the Aβ+ MCI group, MK levels and PiB SUVr correlated positively within sub-regions of the parietal, temporal, and occipital cortices (Figure 2). A single cluster in the right parieto-occipital region remained significant after FWE-correction. Notably, these correlations did not overlap with all regions that showed increased cortical MK (Figure 1) or significantly elevated PiB SUVr (Figure 2) in the group of Aβ+ MCI patients. We observed no significant group differences in MK in any of the assessed subcortical structures.

Figure 2

Figure 2. Statistical t-value maps showing significant (p < 0.05) differences in PiB SUVr and vertex-wise linear correlation between mean kurtosis and PiB SUVr, in amyloid positive (Aβ+) patients with mild cognitive impairment (MCI). Negative t-values indicate decrease in Aβ+ MCIs or negative correlation (blue nuances), while positive t-values indicate increase in Aβ+ MCIs or positive correlation (red nuances). The white outlining marks clusters surviving family-wise error correction for multiple comparisons (p < 0.001). Amyloid status, either positive or amyloid negative, was defined as an atlas based composite regional Pittsburgh compound B (PiB) standard uptake value ratio (SUVR) >1.5 and ≤ 1.5, respectively. The surface-based statistics were adjusted for age using linear regression.

3.2 Mean diffusivity

Cortical MD was trending higher in both Aβ+ and Aβ- MCI patients compared to the CN subjects and in Aβ+ MCI patients compared to Aβ- MCI patients (Supplementary Figures S2, S4). While none of the group differences survived the conservative FWE correction for multiple comparisons, the spatial extent of these changes was largest in the group of Aβ+ MCI patients. Cortical levels of MD showed trends toward correlating negatively with overlapping levels of PiB SUVr in Aβ+ MCIs (Supplementary Figure S2). While weak, these correlations could be the result of Aβ accumulation and/or microglial activation limiting diffusion of water in the tissue, thereby reducing MD. The areas of correlation overlapped with areas of MK–PiB SUVr correlations (Figure 2), but were less widespread (Supplementary Figure S2). This differs from the MK findings indicating that separate mechanisms might affect MD in the patients. We found no significant group differences in MD in any of the assessed subcortical structures.

3.3 Age effect on diffusion-weighted MRI

Levels of MD correlated positively with subject age within parts of the temporal, cingulate, and occipital cortex (Figure 3). Particularly, strong correlations were observed in the hippocampal area, in the lingual and fusiform gyri, in the subgenual anterior cingulate cortex, and in the parieto-occipital- and lateral temporal lobe. In contrast, we observed no significant correlations between MK and age (Figure 3). Age adjustment was made for all group comparisons.

Figure 3

Figure 3. Statistical t-value maps showing significant (p < 0.05) vertex-wise linear correlation between mean kurtosis or mean diffusivity and age for all subjects. Negative t-values indicate negative correlation (blue nuances), while positive t-values indicate positive correlation (red nuances). The white outlining marks clusters surviving family-wise error correction for multiple comparisons (p < 0.001). PiB (Pittsburgh Compound B) status was defined by an atlas based composite regional 11C-PiB SUVr (standard uptake value ratio) >1.5 and ≤ 1.5, respectively. All correlations were adjusted for study group, either mild cognitive impairment (MCI) or cognitively normal, using linear regression.

4 Discussion

Using DKI, we show that microstructural changes to the cortex can be observed as elevated MK in patients with prodromal AD before cortical atrophy appears. Secondly, we show that MD, but not MK, increases considerably with subject age, emphasizing the need to adjust for age when evaluating MD.

We suggest that the microstructural changes detected by MK may reflect the toxic effects of Aβ pathology during the prodromal phase of AD, either as a direct effect of Aβ oligomers or an indirect effect of increased neuroinflammation. Consistent with this interpretation, we found significantly higher MK values in Aβ+ MCI patients than in both Aβ- MCI patients and in CN subjects. These elevations were most pronounced in the temporal cortex and precuneus of Aβ+ MCI patients overlapping with regions typically affected by early deposition of Aβ plaques in AD (Lecy et al., 2024). Within these cortical regions, the most prominent hotspots of elevated MK involved parts of the left lateral temporal lobe, and the caudal part of the right precuneus close to the parietooccipital sulcus. In addition, levels of MK correlated positively with levels of ¹¹C-PiB SUVr, within sub-regions of the parietal and temporal lobe in the Aβ+ MCIs. The FWE-surviving cluster of significant MK–PiB SUVr correlations was just outside the areas with significant levels of PiB SUVr and did not overlap with areas of significant groups differences in MK. Aβ is known to start aggregating long before symptoms appear and, at the time of diagnosis, amyloid has reached a plateau in many of the primary affected regions (Jack et al., 2013). Thus, correlates of amyloid will be confined to areas where Aβ build up is ongoing. Taken together, our findings indicate that MK may be sensitive to build-up of Aβ aggregates in the tissue (Nam et al., 2025). Within the regions with PiB SUVr–MK correlations, MD was found to be negatively correlated with PiB SUVr strengthening the observation, that aggregation of amyloid leads to restriction of water diffusion. Note that these changes are found in the absence of any significant cortical atrophy, indicating that elevated MK precedes the cortical degenerative changes that develop in parallel with cognitive symptoms in subjects with AD (Jack et al., 2013). In support of our findings, elevated MK has been detected in the cortex and thalamus of transgenic mice with established amyloidosis compared with wild type mice (Vanhoutte et al., 2013). This has been corroborated by observations of greatly increased MK in numerous cortical regions with a high degree of Aβ pathology in the APP/PS1 mice and by a positive correlation between MK and levels of anti-Aβ (clone 4G8) antibody in the same mice (Praet et al., 2018). Similar to the Aβ+ MCI patients examined in our study, the mice did not have established neurodegeneration or major cerebral atrophy, despite having widespread Aβ pathology.

The mechanisms driving the observed increases in MK are unclear. Extracellular Aβ plaques introduce numerous impermeable boundaries into brain tissue. Water molecules diffusing in the extracellular space encounter these plaques, which increase the heterogeneity of diffusion paths and may cause deviations from Gaussian diffusion, captured by kurtosis metrics. Reactive gliosis (astrocytes and microglia) and inflammation occur around plaques. These processes increase cellular density and branching (GFAP-positive astrocytes, IBA1-positive microglia) and add further structural barriers and tortuosity to water diffusion. Praet et al. showed strong correlations between MK and markers of gliosis (GFAP, IBA1) and amyloid load (4G8 staining; Praet et al., 2018).

These findings contrast with other studies, which have addressed DKI in MCI patients with established atrophy and found reductions, rather than increases, in MK. Specifically, reduced MK has been identified in the thalamus, putamen, globus pallidus, and the hippocampus of elderly humans (mean age of 75 years) with late stage MCI (group average MMSE = 23.1; Gong et al., 2017), and in the left hippocampus in a group of relatively less cognitively impaired amnestic MCI patients (group mean MMSE = 26.9) with atrophic hippocampi (Wang et al., 2015). In the latter study, atrophy was detected in the region showing reduced MK, while in the first study, the patients included were at an advanced disease state where extensive brain atrophy is common. The reductions in MK in the MCI patients in these two studies could therefore indicate degenerative changes including loss of cellular boundaries or increased extracellular space, rather than Aβ deposits or neuroinflammation, which were not measured. As MK seemingly initially increases with increasing levels of Aβ (Figure 2; Vanhoutte et al., 2013; Praet et al., 2018), we speculate that these studies might have been biased toward subjects affected by degenerative changes more so than by Aβ deposition. This underlines that MK is in fact sensitive to any microstructural alterations that influence water diffusion in tissue. As MCI progresses toward a clinical AD diagnosis and subsequently advanced stages, neurodegeneration increasingly affects brain regions. On the other hand, the cerebral concentration of Aβ has typically reached a plateau before an AD diagnosis is made (Jack et al., 2013). Because increasing levels of Aβ may elevate MK, while neurodegeneration seemingly reduces MK, MK values may ultimately decrease as the disease advances and neuronal loss worsens. Corroboratively, reduced MK has been observed in patients with moderate to severe AD compared with cognitively normal controls bilaterally in the hippocampus, while mild AD cases showed no MK reductions despite significant atrophy in several brain regions (Wang et al., 2018).

Our results show that MK is a stronger candidate for detection of early microstructural changes in prodromal Alzheimer's disease than MD. Although, elevated cortical MD was observed in both groups of Aβ+ and Aβ- MCI patients compared with cognitively normal controls, MD only weakly correlated with levels of Aβ and in regions overlapping with areas of significant MK–Aβ correlation (Supplementary Figure S2 and Figure 2). Our results point toward changes in MD as mainly an effect of aging (Figure 3), and are thus etiologically non-specific (Muller et al., 2007). The MD changes observed in our study, although weaker, follow in line with previous observations of elevated MD in the inferior temporal lobe of MCI patients (Gyebnar et al., 2018), and in the parietal lobe of MCI patients with more advanced cognitive symptoms (Gong et al., 2017).

In our ROI analysis of hippocampus, amygdala, thalamus, nucleus accumbens, nucleus caudatus, putamen, and globus pallidus, we did not observe any significant subcortical GM changes in either MK or MD between MCI patients and CN subjects. This accords with previous findings that suggest subcortical microstructural changes only emerge at the late stages of MCI and when patients develop clinical AD (Cherubini et al., 2010; Gong et al., 2017). It should be noted that there is a technical limitation to evaluating subcortical microstructural changes with diffusion MRI, since signal-to-noise ratios decrease with distance from the RF coil (Kwok, 2022).

Our interpretations of the data should be taken with caution. First, the limited sample size warrants replications in larger cohorts. Second, the exact mechanism behind the increased MK is unknown. We consider accumulation of proteins and breakdown of membranes as the primary sources of changes to the diffusion parameters measured. However, other factors, such as edema or gliosis, which may originate from a variety of neuropathological conditions, may contribute as well. Third, in almost half of the cognitively normal group amyloid status was not determined with ¹¹C-PiB-PET scans. Finally, the sensitivity of the fast DKI sequence can be questioned as only a few diffusion directions are used for fitting the kurtosis model, which may also explain why no subcortical alterations were detected.

Future studies should address whether a DKI sequence with more diffusion directions increases the sensitivity of MK at the expense of increase in scan time. Moreover, a longitudinal study design should be applied to map the temporal trajectory of MK and to address whether MK obtained with fast DKI can be used to identify subjects with prodromal AD.

In conclusion, we found elevated MK in Aβ-positive MCI patients compared to Aβ-negative MCI patients and significant PiB SUVr–MK correlations suggesting that DKI can detect early microstructural changes linked to amyloid pathology before the onset of cortical atrophy.

Data availability statement

The raw data supporting the conclusions of this article is available here: https://doi.org/10.6084/m9.figshare.30850301.

Ethics statement

The studies involving humans were approved by the Regional Ethics Committee for Biomedical Research in the Central Denmark Region. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

Author contributions

RN: Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing – original draft. PP: Conceptualization, Data curation, Investigation, Methodology, Project administration, Writing – review & editing. RI: Conceptualization, Data curation, Investigation, Methodology, Project administration, Writing – review & editing. RD: Writing – review & editing. AT: Data curation, Validation, Writing – review & editing. HB: Conceptualization, Data curation, Investigation, Resources, Writing – review & editing. HG: Data curation, Investigation, Resources, Writing – review & editing. DB: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – review & editing. LØ: Conceptualization, Funding acquisition, Investigation, Project administration, Resources, Supervision, Writing – review & editing. SE: Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Writing – original draft, Writing – review & editing.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This study was supported by research grants from The Danish Research Council for Independent Research, grant agreement numbers DFF-1331-00184 and DFF-4004-00305, and from the Lundbeck Foundation (grant no. R140-2013-13245). LØ received funding from the VELUX Foundation (Arcadia – Aarhus Research Center for Aging and Dementia) and from the EU Joint Programming initiative within Neurodegenerative Diseases, funded by the Danish Strategic Research Council (APGeM - Pre-clinical genotype-phenotype predictors of Alzheimer's disease and other dementias, grant agreement number 3056-00001).

Acknowledgments

We thank radiographers Dora Grauballe and Michael Geneser for help with acquiring MRI, and Irene Klærke Mikkelsen for providing technical support during data processing.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The author SE declared that they were an editorial board member of Frontiers at the time of submission. This had no impact on the peer review process and the final decision.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

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

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/frdem.2025.1725754/full#supplementary-material

Abbreviations

AD, Alzheimer's disease; Aβ, β-amyloid-42; Aβ+, amyloid-positive; Aβ−, amyloid-negative; CDR, clinical dementia rating; DKI, diffusion kurtosis imaging; DWI, diffusion weighted imaging; GDS-15, geriatric depression scale; GM, gray matter; CN, cognitively normal; MCI, mild cognitive impairment; MD, mean diffusivity; MK, mean kurtosis; MMSE, mini-mental state examination; MRI, magnetic resonance imaging; PET, positron emission tomography; PiB, Pittsburgh compound-B; SUVr, standard uptake value ratio; T₂w-FLAIR, T₂-weighted fluid attenuated inversion recovery; WM, White matter; WMH, White matter hyperintensity.

References

Arab, A., Wojna-Pelczar, A., Khairnar, A., and Szabo, N.. (2018). Principles of diffusion kurtosis imaging and its role in early diagnosis of neurodegenerative disorders. Brain Res. Bull. 139, 91–98. doi: 10.1016/j.brainresbull.2018.01.015

PubMed Abstract | Crossref Full Text | Google Scholar

Aubert-Broche, B., Fonov, V. S., Garcia-Lorenzo, D., Mouiha, A., Guizard, N., Coupe, P., et al. (2013). A new method for structural volume analysis of longitudinal brain MRI data and its application in studying the growth trajectories of anatomical brain structures in childhood. Neuroimage 82, 393–402. doi: 10.1016/j.neuroimage.2013.05.065

PubMed Abstract | Crossref Full Text | Google Scholar

Avants, B. B., Tustison, N. J., Song, G., Cook, P. A., Klein, A., and Gee, J. C. (2011). A reproducible evaluation of ANTs similarity metric performance in brain image registration. Neuroimage 54, 2033–2044. doi: 10.1016/j.neuroimage.2010.09.025

PubMed Abstract | Crossref Full Text | Google Scholar

Basser, P. J., and Pierpaoli, C. (1996). Microstructural and physiological features of tissues elucidated by quantitative-diffusion-tensor MRI. J. Magn. Reson. B 111, 209–219. doi: 10.1006/jmrb.1996.0086

PubMed Abstract | Crossref Full Text | Google Scholar

Cherubini, A., Peran, P., Spoletini, I., Di Paola, M., Di Iulio, F., Hagberg, G. E., et al. (2010). Combined volumetry and DTI in subcortical structures of mild cognitive impairment and Alzheimer's disease patients. J. Alzheimers Dis. 19, 1273–1282. doi: 10.3233/JAD-2010-091186

PubMed Abstract | Crossref Full Text | Google Scholar

Collins, D. L., Neelin, P., Peters, T. M., and Evans, A. C. (1994). Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space. J. Comput. Assist. Tomogr. 18, 192–205. doi: 10.1097/00004728-199403000-00005

PubMed Abstract | Crossref Full Text | Google Scholar

Coupe, P., Manjon, J. V., Fonov, V., Pruessner, J., Robles, M., and Collins, D. L. (2011). Patch-based segmentation using expert priors: application to hippocampus and ventricle segmentation. NeuroImage 54, 940–954. doi: 10.1016/j.neuroimage.2010.09.018

PubMed Abstract | Crossref Full Text | Google Scholar

Eskildsen, S. F., Coupe, P., Fonov, V., Manjon, J. V., Leung, K. K., Guizard, N., et al. (2012). BEaST: brain extraction based on nonlocal segmentation technique. NeuroImage 59, 2362–2373. doi: 10.1016/j.neuroimage.2011.09.012

PubMed Abstract | Crossref Full Text | Google Scholar

Eskildsen, S. F., Gyldensted, L., Nagenthiraja, K., Nielsen, R. B., Hansen, M. B., Dalby, R. B., et al. (2017). Increased cortical capillary transit time heterogeneity in Alzheimer's disease: a DSC-MRI perfusion study. Neurobiol. Aging 50, 107–118. doi: 10.1016/j.neurobiolaging.2016.11.004

PubMed Abstract | Crossref Full Text | Google Scholar

Eskildsen, S. F., and Ostergaard, L. R. (2006). Active surface approach for extraction of the human cerebral cortex from MRI. Med. Image Comput. Comput. Assist. Interv. 9, 823–830. doi: 10.1007/11866763_101

PubMed Abstract | Crossref Full Text | Google Scholar

Falangola, M. F., Jensen, J. H., Babb, J. S., Hu, C., Castellanos, F. X., Di Martino, A., et al. (2008). Age-related non-Gaussian diffusion patterns in the prefrontal brain. J. Magn. Reson. Imaging 28, 1345–1350. doi: 10.1002/jmri.21604

PubMed Abstract | Crossref Full Text | Google Scholar

Frisoni, G. B., Fox, N. C., Scheltens, P., and Thompson, P. M. (2010). The clinical use of structural MRI in Alzheimer disease. Nat. Rev. Neurol. 6, 67–77. doi: 10.1038/nrneurol.2009.215

PubMed Abstract | Crossref Full Text | Google Scholar

Gong, N. J., Chan, C. C., Leung, L. M., Wong, C. S., Dibb, R., and Liu, C. (2017). Differential microstructural and morphological abnormalities in mild cognitive impairment and Alzheimer's disease: evidence from cortical and deep gray matter. Hum. Brain Mapp. 38, 2495–2508. doi: 10.1002/hbm.23535

PubMed Abstract | Crossref Full Text | Google Scholar

Gong, N. –J., Wong, C. -S., Chan, C. -C., Leung, L. -M., and Chu, Y. -C. (2014). Aging in deep gray matter and white matter revealed by diffusional kurtosis imaging. Neurobiol. Aging 35, 2203–2216. doi: 10.1016/j.neurobiolaging.2014.03.011

PubMed Abstract | Crossref Full Text | Google Scholar

Gyebnar, G., Szabo, A., Siraly, E., Fodor, Z., Sakovics, A., Salacz, P., et al. (2018). What can DTI tell about early cognitive impairment?—Differentiation between MCI subtypes and healthy controls by diffusion tensor imaging. Psychiatry Res. Neuroimaging 272, 46–57. doi: 10.1016/j.pscychresns.2017.10.007

Crossref Full Text | Google Scholar

Hansen, B., Lund, T. E., Sangill, R., and Jespersen, S. N. (2013). Experimentally and computationally fast method for estimation of a mean kurtosis. Magn. Reson. Med. 69, 1754–1760. doi: 10.1002/mrm.24743

PubMed Abstract | Crossref Full Text | Google Scholar

Hansen, B., Lund, T. E., Sangill, R., and Jespersen, S. N. (2014). Experimentally and computationally fast method for estimation of a mean kurtosis. Magn. Reson. Med. 71, 2250–2250. doi: 10.1002/mrm.25090

PubMed Abstract | Crossref Full Text | Google Scholar

Jack, C. R. Jr., Knopman, D. S., Jagust, W. J., Petersen, R. C., Weiner, M. W., Aisen, P. S., et al. (2013). Tracking pathophysiological processes in Alzheimer's disease: an updated hypothetical model of dynamic biomarkers. Lancet Neurol. 12, 207–216. doi: 10.1016/S1474-4422(12)70291-0

PubMed Abstract | Crossref Full Text | Google Scholar

Jensen, J. H., and Helpern, J. A. (2010). MRI quantification of non-Gaussian water diffusion by kurtosis analysis. NMR Biomed. 23, 698–710. doi: 10.1002/nbm.1518

PubMed Abstract | Crossref Full Text | Google Scholar

Jespersen, S. N., Bjarkam, C. R., Nyengaard, J. R., Chakravarty, M. M., Hansen, B., Vosegaard, T., et al. (2010). Neurite density from magnetic resonance diffusion measurements at ultrahigh field: comparison with light microscopy and electron microscopy. NeuroImage 49, 205–216. doi: 10.1016/j.neuroimage.2009.08.053

PubMed Abstract | Crossref Full Text | Google Scholar

Kale, R. A., Gupta, R. K., Saraswat, V. A., Hasan, K. M., Trivedi, R., Mishra, A. M., et al. (2006). Demonstration of interstitial cerebral edema with diffusion tensor MR imaging in type C hepatic encephalopathy. Hepatology 43, 698–706. doi: 10.1002/hep.21114

PubMed Abstract | Crossref Full Text | Google Scholar

Kellner, E., Dhital, B., Kiselev, V. G., and Reisert, M. (2016). Gibbs-ringing artifact removal based on local subvoxel-shifts. Magn. Reson. Med. 76, 1574–1581. doi: 10.1002/mrm.26054

PubMed Abstract | Crossref Full Text | Google Scholar

Kwok, W. E. (2022). Basic principles of and practical guide to clinical MRI radiofrequency coils. Radiographics 42, 898–918. doi: 10.1148/rg.210110

PubMed Abstract | Crossref Full Text | Google Scholar

Lawrence, K. E., Nabulsi, L., Santhalingam, V., Abaryan, Z., Villalon-Reina, J. E., Nir, T. M., et al. (2021). Age and sex effects on advanced white matter microstructure measures in 15,628 older adults: a UK biobank study. Brain Imaging Behav. 15, 2813–2823. doi: 10.1007/s11682-021-00548-y

PubMed Abstract | Crossref Full Text | Google Scholar

Lecy, E. E., Min, H. -K., Apgar, C. J., Maltais, D. D., Lundt, E. S., Albertson, S. M., et al. (2024). Patterns of early neocortical amyloid-β accumulation: a PET population-based study. J. Nucl. Med. 65, 1122–1128. doi: 10.2967/jnumed.123.267150

PubMed Abstract | Crossref Full Text | Google Scholar

Lee, P., Ryoo, H., Park, J., and Jeong, Y. (2017). Morphological and microstructural changes of the hippocampus in early MCI: a study utilizing the Alzheimer's disease neuroimaging initiative database. J. Clin. Neurol. 13, 144–154. doi: 10.3988/jcn.2017.13.2.144

PubMed Abstract | Crossref Full Text | Google Scholar

Marques, J. P., Kober, T., Krueger, G., van der Zwaag, W., Van de Moortele, P. F., and Gruetter, R. (2010). MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. NeuroImage 49, 1271–1281. doi: 10.1016/j.neuroimage.2009.10.002

PubMed Abstract | Crossref Full Text | Google Scholar

Matsumoto, N., Sugimoto, T., Yamashita, F., Mori, F., Kuroda, Y., Fujita, K., et al. (2025). A diffusion kurtosis imaging study of the relationship between whole brain microstructure and cognitive function in older adults with mild cognitive impairment. Acta Radiol. 66, 107–114. doi: 10.1177/02841851241295394

PubMed Abstract | Crossref Full Text | Google Scholar

Muller, M. J., Greverus, D., Weibrich, C., Dellani, P. R., Scheurich, A., Stoeter, P., et al. (2007). Diagnostic utility of hippocampal size and mean diffusivity in amnestic MCI. Neurobiol. Aging 28, 398–403. doi: 10.1016/j.neurobiolaging.2006.01.009

PubMed Abstract | Crossref Full Text | Google Scholar

Nam, Y., Shin, S. J., Kumar, V., Won, J., Kim, S., and Moon, M. (2025). Dual modulation of amyloid beta and tau aggregation and dissociation in Alzheimer's disease: a comprehensive review of the characteristics and therapeutic strategies. Transl. Neurodegener 14:15. doi: 10.1186/s40035-025-00479-4

PubMed Abstract | Crossref Full Text | Google Scholar

Nielsen, R. B., Egefjord, L., Angleys, H., Mouridsen, K., Gejl, M., Moller, A., et al. (2017). Capillary dysfunction is associated with symptom severity and neurodegeneration in Alzheimer's disease. Alzheimers Dement 13, 1143–1153. doi: 10.1016/j.jalz.2017.02.007

PubMed Abstract | Crossref Full Text | Google Scholar

Parbo, P., Ismail, R., Hansen, K. V., Amidi, A., Marup, F. H., Gottrup, H., et al. (2017). Brain inflammation accompanies amyloid in the majority of mild cognitive impairment cases due to Alzheimer's disease. Brain J. Neurol. 140, 2002–2011. doi: 10.1093/brain/awx120

PubMed Abstract | Crossref Full Text | Google Scholar

Petersen, R. C. (2004). Mild cognitive impairment as a diagnostic entity. J. Intern. Med. 256, 183–194. doi: 10.1111/j.1365-2796.2004.01388.x

PubMed Abstract | Crossref Full Text | Google Scholar

Praet, J., Manyakov, N. V., Muchene, L., Mai, Z., Terzopoulos, V., de Backer, S., et al. (2018). Diffusion kurtosis imaging allows the early detection and longitudinal follow-up of amyloid-beta-induced pathology. Alzheimers Res. Ther. 10:1. doi: 10.1186/s13195-017-0329-8

Crossref Full Text | Google Scholar

Schmidt, P., Gaser, C., Arsic, M., Buck, D., Forschler, A., Berthele, A., et al. (2012). An automated tool for detection of FLAIR-hyperintense white-matter lesions in multiple sclerosis. Neuroimage 59, 3774–3783. doi: 10.1016/j.neuroimage.2011.11.032

PubMed Abstract | Crossref Full Text | Google Scholar

Sexton, C. E., Kalu, U. G., Filippini, N., Mackay, C. E., and Ebmeier, K. P. (2011). A meta-analysis of diffusion tensor imaging in mild cognitive impairment and Alzheimer's disease. Neurobiol. Aging 32, 2322 e2325–e2318. doi: 10.1016/j.neurobiolaging.2010.05.019

PubMed Abstract | Crossref Full Text | Google Scholar

Smart, S. D., Firbank, M. J., and O'Brien, J. T. (2011). Validation of automated white matter hyperintensity segmentation. J. Aging Res. 2011:391783. doi: 10.4061/2011/391783

PubMed Abstract | Crossref Full Text | Google Scholar

Struyfs, H., Van Hecke, W., Veraart, J., Sijbers, J., Slaets, S., De Belder, M., et al. (2015). Diffusion kurtosis imaging: a possible MRI biomarker for AD diagnosis? J. Alzheimers Dis. 48, 937–948. doi: 10.3233/JAD-150253

PubMed Abstract | Crossref Full Text | Google Scholar

Taha, H. T., Chad, J. A., and Chen, J. J. (2022). DKI enhances the sensitivity and interpretability of age-related DTI patterns in the white matter of UK biobank participants. Neurobiol. Aging 115, 39–49. doi: 10.1016/j.neurobiolaging.2022.03.008

PubMed Abstract | Crossref Full Text | Google Scholar

Taoka, T., Yasuno, F., Morikawa, M., Inoue, M., Kiuchi, K., Kitamura, S., et al. (2016). Diffusion tensor studies and voxel-based morphometry of the temporal lobe to determine the cognitive prognosis in cases of Alzheimer's disease and mild cognitive impairment: Do white matter changes precede gray matter changes? Springerplus 5, 1023. doi: 10.1186/s40064-016-2692-5

PubMed Abstract | Crossref Full Text | Google Scholar

Vanhoutte, G., Pereson, S., Delgado, Y. P. R., Guns, P. J., Asselbergh, B., Veraart, J., et al. (2013). Diffusion kurtosis imaging to detect amyloidosis in an APP/PS1 mouse model for Alzheimer's disease. Magn. Reson. Med. 69, 1115–1121. doi: 10.1002/mrm.24680

PubMed Abstract | Crossref Full Text | Google Scholar

Veraart, J., Novikov, D. S., Christiaens, D., Ades-Aron, B., Sijbers, J., and Fieremans, E. (2016). Denoising of diffusion MRI using random matrix theory. Neuroimage 142, 394–406. doi: 10.1016/j.neuroimage.2016.08.016

PubMed Abstract | Crossref Full Text | Google Scholar

Wang, D., Guo, Z. H., Liu, X. H., Li, Y. H., and Wang, H. (2015). Examination of hippocampal differences between Alzheimer disease, amnestic mild cognitive impairment and normal aging: diffusion kurtosis. Curr. Alzheimer Res. 12, 80–87. doi: 10.2174/1567205012666141218142422

PubMed Abstract | Crossref Full Text | Google Scholar

Wang, M. L., Wei, X. E., Fu, J. L., Li, W., Yu, M. M., Li, P. Y., et al. (2018). Subcortical nuclei in Alzheimer's disease: a volumetric and diffusion kurtosis imaging study. Acta Radiol 59, 1365–1371. doi: 10.1177/0284185118758122

PubMed Abstract | Crossref Full Text | Google Scholar

Weston, P. S., Simpson, I. J., Ryan, N. S., Ourselin, S., and Fox, N. C. (2015). Diffusion imaging changes in grey matter in Alzheimer's disease: a potential marker of early neurodegeneration. Alzheimers Res. Ther. 7:47. doi: 10.1186/s13195-015-0132-3

PubMed Abstract | Crossref Full Text | Google Scholar

Worsley, K. J., Marrett, S., Neelin, P., Vandal, A. C., Friston, K. J., and Evans, A. C. (1996). A unified statistical approach for determining significant signals in images of cerebral activation. Hum. Brain Mapp. 4, 58–73. doi: 10.1002/(SICI)1097-0193(1997)5:2<79::AID-HBM1>3.0.CO;2-J

PubMed Abstract | Crossref Full Text | Google Scholar