https://orcid.org/0000-0001-6645-9162
Figures
Abstract
Diffusion magnetic resonance imaging (dMRI) is a non-invasive neuroimaging technique that enables in vivo assessment of white matter microstructure and is highly sensitive to tissue alterations associated with disease. Although substantial evidence links diffusion-derived metrics to underlying white matter tissue properties, the presence of complex within-voxel axonal configurations complicates their biological interpretation. Several methods have been proposed to assess diffusion properties of individual crossing axonal populations, but their validation and clinical applicability remain limited. Glaucoma, the second leading cause of blindness worldwide, is characterized by progressive loss of retinal ganglion cells and axonal damage in the optic nerve, leading to degeneration along the entire visual pathway. This degeneration includes secondary effects on fiber crossings within the optic chiasm, which are challenging to characterize with conventional diffusion methods. Here, we evaluated whether advanced diffusion metrics can detect microstructural alterations in these complex white matter configurations and whether these measures correlate with clinical markers of glaucoma severity. In this study, we evaluated 31 patients with asymmetric glaucoma and 31 healthy controls using advanced diffusion magnetic resonance imaging methods, including Diffusion Tensor Imaging, Constrained Spherical Deconvolution, multi-tensor fit via Multi-Resolution Discrete Search method, and Fixel-Based Analysis. We found significant differences of diffusion metrics in white matter tracts of the visual system, including the optic nerve, optic chiasm, optic tracts, and optic radiations. Moreover, diffusion metrics correlated with clinical ophthalmological parameters such as cup-to-disc ratio, visual field mean deviation, and retinal nerve fiber layer thickness. These findings support the use of advanced diffusion magnetic resonance imaging models as sensitive tools for detecting Wallerian degeneration and resolving complex white matter architecture in the human visual pathway, and demonstrate their utility to study other fiber-crossing regions throughout the brain.
Citation: Coutiño D, Guerrero-Zavala J, Domínguez-Frausto CA, García-Guillén M, Coronado-Leija R, Ramírez-Manzanares A, et al. (2026) Disentangling crossing fibers with advanced dMRI methods reveals bundle-specific degeneration across the visual system in asymmetric glaucoma. PLoS One 21(6): e0349951. https://doi.org/10.1371/journal.pone.0349951
Editor: Md Nasir Uddin, University of Rochester, UNITED STATES OF AMERICA
Received: November 11, 2025; Accepted: May 7, 2026; Published: June 22, 2026
Copyright: © 2026 Coutiño et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All anonymized images and patient information are available in BIDS format at https://openneuro.org/datasets/ds006837.
Funding: Funding was provided by Dirección General de Asuntos del Personal Académico, Universidad Nacional Autónoma de México (PAPIIT IN213423) and Secretaría de Ciencia, Humanidades, Tecnología e Innovación (CF-2023-I-218). DC is a doctoral student from Programa de Doctorado en Ciencias Biomédicas, Universidad Nacional Autónoma de México (UNAM) and has recieved a fellowship (CVU 1288997) from Secretaría de Ciencia, Humanidades, Tecnología e Innovación (SECIHTI, formerly CONAHCYT). The National Laboratory for MRI has received funding from Secretaría de Ciencia, Humanidades, Tecnología e Innovación. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Diffusion-weighted magnetic resonance imaging (dMRI) enables noninvasive assessment of tissue microstructure [1–3]. Diffusion tensor imaging (DTI) [4] is the most commonly used method, given its efficiency and wide availability. However, DTI represents diffusion in only one prominent direction, rendering it unsuitable for regions with complex fiber configurations, such as fiber crossings—which are present in the majority of human white matter [5–7]. This limitation causes the inaccurate estimation of oblate or spherical tensors in regions containing multiple fiber orientations, thereby reducing fractional anisotropy (FA). Moreover, in cases when one of the crossing fiber populations is degenerated, the remaining bundle dominates the diffusion signal, and the resulting tensor appears to be anisotropic (i.e., with increased FA), leading to incorrect biological interpretations [8].
Other methods for analyzing the diffusion signal can resolve multiple fiber orientations. Independent estimation of diffusion tensors for each fiber bundle within each voxel can be achieved with Multi-Resolution Discrete Search (MRDS), a method with relatively low data acquisition requirements and robustness to noise [9]. Similarly, Constrained Spherical Deconvolution (CSD) [10] can identify one or more fiber elements (fixels) and derive apparent fiber density (AFD) values independently for each fiber bundle [11]. In previous work using an animal model of unilateral retinal ischemia and high-resolution ex vivo dMRI, our group showed that multi-tensor approaches, as well as CSD analyses, are able to correctly identify a known region of fiber crossings in the optic chiasm and, importantly, provide independent and accurate microstructural information related to the degenerated and intact axonal populations that cross within this structure [12]. Moreover, the per-bundle diffusion metrics derived from MRDS and CSD were tightly correlated with the degree of degeneration of the optic nerves assessed with electron microscopy. The ability to disentangle intact and degenerated fiber populations through bundle-wise analyses has immediate applications to characterize other regions of white matter with complex configurations [13], yet they require further validation when derived from dMRI acquisitions with parameters constrained to clinical settings (i.e., relatively short acquisition times and lower resolution). Therefore, an important next step is to evaluate whether these approaches remain sensitive to microstructural alterations when applied to in vivo human data acquired under standard clinical limitations.
In this work, asymmetric glaucoma, characterized by unequal severity of damage between the two eyes, serves as a case study that allows for the spatial assessment of diffusion metrics throughout the visual pathway. Glaucoma is a leading cause of irreversible blindness worldwide that affects 3.54% of the population aged 40–80 years [14]. It causes progressive degeneration of the retinal ganglion cells and leads to gradual peripheral vision loss [15]. Generally, by the time of diagnosis, 30–50% of retinal ganglion cells have already been lost. The current evaluation of patients with glaucoma relies on visual field parameters such as mean deviation (VF-MD) and on optical coherence tomography (OCT) metrics, including the vertical cup-to-disc ratio (vCD) and retinal nerve fiber layer (RNFL) thickness, which reflect the degree of retinal degeneration. The axons that extend from retinal ganglion cells form the optic nerve, and suffer Wallerian degeneration upon damage to their soma. Several studies have shown correlations between diffusion metrics of the optic pathway and said clinical parameters [16–19]. However, previous reports have avoided analyzing the optic chiasm due to its complex crossing fibers. Because the optic chiasm contains two anatomically distinct fiber populations that partially decussate, it provides a particularly informative model to evaluate the performance of diffusion methods in regions with known crossing-fiber architecture.
This study investigates the potential of advanced dMRI to detect in vivo structural alterations of the optic pathway in patients with asymmetric glaucoma. Importantly, the use of asymmetric glaucoma in this study is not only clinically relevant, but also provides a unique experimental framework to evaluate diffusion methods in a region with known crossing-fiber architecture. We focus on the optic chiasm as a model of crossing-fiber architecture, and the asymmetric pattern of retinal damage caused by glaucoma provides a clear scenario in which two fiber populations with differing degrees of degeneration cross in a known anatomical structure. Our objectives are (i) to validate per-bundle dMRI analyses in clinically-accessible dMRI data, using the optic chiasm as a biologically grounded model of fiber crossing, with the goal that such validation may extend to other white-matter regions with crossing fibers throughout the brain and (ii) to characterize white matter damage throughout the entire visual pathway and assess the correlation of diffusion metrics with clinical parameters.
Materials and methods
Participants
This prospective case–control study included patients with asymmetric glaucoma and healthy control subjects. We recruited adult patients with asymmetric glaucoma from the Glaucoma Service at the Mexican Institute of Ophthalmology (IMO). A total of 1,926 clinical records of patients evaluated in the last 12 months were screened. The majority of patients were excluded due to symmetric glaucoma (802), age outside the inclusion range (654), single functioning eye (158), incomplete diagnostic studies (94), or other clinical and radiological contraindications, such as the presence of pacemakers, metallic implants, or mobility aids incompatible with MRI (12). Despite potential eligibility, 171 patients were not willing to participate or were not found based on their contact information. After applying these criteria, 35 patients underwent imaging. Four patients were excluded due to unexpected radiological findings. This resulted in a final cohort of 31 patients (PX) for analysis (14 females and 17 males) with age 59 ± 14 years old (mean ± SD). Patients were considered to have clearly asymmetric glaucoma when there was an interocular difference of their vertical cup-to-disc ratio (vCD) , together with an interocular difference in retinal nerve fiber layer (RNFL) thickness
µm on OCT [20,21]. The distribution of glaucoma subtypes was as follows: 18 patients with open-angle glaucoma (OAG), including primary open-angle glaucoma (POAG), 11 with angle-closure glaucoma (ACG), 1 with normal-tension glaucoma (NTG) and 1 with traumatic glaucoma.
Thirty-one controls subjects (CTRL) were recruited to match the age (55 ± 8 years old) and sex distribution (15 females and 16 males) of the patient group. Control participants were included if they had no history of ophthalmological or neurological disorders and were excluded if they had a diagnosis of neurodegenerative disease, visual disorders, or any contraindications to MRI. Clinical variables such as diabetes and hypertension were recorded for all participants; however, given the heterogeneity, these variables were not included as covariates in the statistical analyses. Demographic data from all participants of this study is available in the Supporting Information (S1 Table).
Participation in this study was voluntary and independent from treatment. All procedures were conducted in accordance with the Declaration of Helsinki and participants provided written consent. The protocol was approved by the Research Ethics Committee of the Institute of Neurobiology (019.H-RM) and by the Bioethics Committee of the IMO (CI/IMO-001/2023). Recruitment of participants was performed between April 2023 and March 2025. All participants provided written consent.
Imaging
The dataset was acquired on a 3 T Discovery MR750 scanner (GE Healthcare, Milwaukee, WI, USA) with a 32-channel head coil. We acquired multi-shell dMRI using two different methods, both had three non-zero b-shells (b = 800, 1500, 2500 s/mm²) with diffusion sensitization in 16, 32, and 64 different directions. Six b = 0 s/mm² volumes were also acquired. A Multiplexed sensitivity-encoding (MUSE) dMRI sequence [22] was used to acquire an axial slab encompassing the visual pathway and oriented parallel to the optic nerves (3 shots, 28 slices, resolution of 1.7×1.5×1.7 mm³ interpolated by the scanner to 0.86×0.86 in-plane resolution, TR = 3500 ms, TE = 80 ms). In contrast to conventional single-shot EPI acquisitions, multi-shot approaches sample k-space across multiple excitations, which reduces susceptibility-related distortions. This is particularly advantageous for imaging the optic nerves and chiasm, which are located near the skull base and are therefore prone to geometric distortions and signal loss in standard diffusion acquisitions, but require longer acquisitions [23,24]. To ensure the inclusion of the optic radiations, whole-brain dMRI were acquired with a simultaneous multi-slice sequence (multiband factor = 2, TR = 4000 ms, TE = 80 ms, isotropic resolution of 1.7 mm). Simultaneous multi-slice acquisitions enable faster whole-brain coverage by exciting multiple slices within a single shot, at the cost of increased sensitivity to susceptibility distortions compared with multi-shot acquisitions. Supplementary S1 Fig shows a visual comparison of the two dMRI acquisitions. Additionally, T1-weighted images (0.8 mm isotropic resolution) and T2-weighted images (1 mm isotropic resolution) were acquired for screening purposes. Total acquisition time was approximately 40 minutes.
Preprocessing
All dMRI data sets were first denoised using either Marchenko–Pastur principal component analysis (MP-PCA) [25,26] for the whole-brain data sets, or Patch2Self [27] for the in-plane interpolated MUSE acquisition. The latter was chosen given the in-plane interpolation performed by the scanner on the MUSE images, which alters the noise structure present in the images that negatively affects the performance of MP-PCA, but not Patch2Self. Gibbs ringing artifacts were removed using the subvoxel-shifts method [28]. Subject motion, as well as magnetic susceptibility and eddy-current induced distortions, were corrected with topup and Eddy tools, available in the FSL toolbox (version 6.0.7.4) [29]. Bias field inhomogeneities were then corrected using the N4 algorithm [30].
Voxel-wise analysis of dMRI
Diffusion tensor and kurtosis imaging (DTI and DKI, respectively) were estimated using DIPY [31,32]. The diffusion tensor yielded parameters of fractional anisotropy (FA) and mean diffusivity (MD). The diffusion kurtosis measures the non-Gaussianity of water diffusion and provides additional information on tissue complexity [33]. Mean kurtosis (MK), axial kurtosis (AK), and radial kurtosis (RK) were calculated at each voxel. Regions of interest (ROI) were manually drawn outlining each optic nerve using mrview (MRtrix 3.0.4) [34], and average DTI and DKI metrics were obtained for each ROI. Manual segmentation was performed by a single rater (DC) who was blinded to participant diagnosis and group status. Visual quality control was performed during ROI delineation and after extraction of diffusion metrics to ensure accurate segmentation and absence of processing artifacts.
Per-bundle analysis of dMRI
To disentangle diffusion metrics in regions of fiber crossings, we applied two distinct and complementary methods that provide practical frameworks for resolving multiple fiber orientations and per-bundle diffusion metrics. Both of these methods have relatively low acquisition requirements and are straightforward to implement on most modern clinical scanners. Per-bundle analyses allowed for the individual assessment of axons originating in the eye most affected by glaucoma (termed here ipsilateral fibers), or in the other eye (contralateral fibers), even in regions where the two populations meet (i.e., the optic chiasm).
Constrained Spherical Deconvolution (CSD)
This method was used to calculate the apparent fiber density (AFD) [11] from the MUSE acquisition, which provided finer detail and minimized geometric distortions around the optic chiasm and nerves. White matter response functions were estimated using a recursive calibration [35], applied within a whole-brain mask. As this analysis was restricted to the optic nerves and chiasm, single-tissue CSD was applied to obtain the white matter fiber orientation distributions (FODs). The resulting FODs were normalized to allow inter-subject comparison. These normalized FODs were used to generate tractography, encompassing fibers from both optic nerves, the optic chiasm, and the initial portion of the optic tracts, as shown in Fig 1. FODs were converted into fixels, restricting the maximum number to two fixels per voxel (i.e., expecting one fixel at the level of optic nerves and tracts, and two fixels within the chiasm), and fixel-AFD maps were derived within manually-defined ROIs in the optic nerves and chiasm delineated by the same rater and under blinded conditions, consistent with the voxel-wise analyses.
Panels A and D show whole-brain T1-weighted images with inverted grayscale and tractography of the visual system overlaid. Panels B and E display zoomed views of the optic chiasm, highlighting the crossing of nasal (contralateral, green) and temporal (ipsilateral, red) fibers. Panels C and F provide schematic representations of a voxel at the fiber crossing. In the control subject, both multi-resolution discrete search (MRDS) and constrained spherical deconvolution (CSD) depict symmetric fiber populations. In the patient, MRDS shows reduced fixel-FA and increased fixel-MD in the ipsilateral fibers (red colored fixels), while CSD demonstrates reduced apparent fiber density (AFD). These alterations illustrate the sensitivity of bundle-specific metrics to detect glaucoma-related degeneration in regions with complex fiber architecture.
Multi-Resolution Discrete Search (MRDS)
A region of interest (ROI) was manually drawn by the same rater to delineate each optic chiasm, where one or two diffusion tensors were fit in each voxel, then used to derive maps of fixel-FA and fixel-MD corresponding to the ipsilateral and contralateral sides of the eye affected by glaucoma. Although MRDS includes different model selectors to determine the number of tensors, in this study, we employed the Track Orientation Density Imaging (TODI) method as the selection strategy [36]. This approach uses tractography obtained with CSD to generate voxel-wise track orientation distributions, which provides a more consistent representation of fiber orientation than FODs [37]. The TODI was used to derive a Number of Fiber Orientations (NuFO) map, which was constrained to , corresponding to the two main fiber populations in the chiasm (one from each optic nerve) [38]. Next, the diffusion tensors best aligned with each optic nerve (either ipsilateral or contralateral to the most affected eye) were selected, from which we derived FA and MD (referred to as fixel-FA and fixel-MD, respectively, to disambiguate these from the standard DTI metrics obtained in the optic nerves). The two metrics were separately averaged for each orientation (i.e., ipsilateral and contralateral to the most affected eye).
Fixel-based analysis (FBA)
Prior to further analysis, all diffusion images were visually inspected to identify potential artifacts that could affect the results. One subject was excluded from the fixel-based analysis due to poor quality of the whole-brain acquisition. The dMRI obtained with simultaneous multi-slice acquisition providing whole-brain coverage was resampled to 1.25 mm isotropic resolution after pre-processing. Response functions for white matter, gray matter, and cerebrospinal fluid were estimated using the unsupervised approach described by Dhollander [39,40], followed by multi-tissue CSD (MSMT-CSD) to compute FODs [41]. Individual FOD images were normalized and a population FOD template was generated using nonlinear registration [42], to which all subjects were registered. From this template, fixel-wise metrics of fiber density (FD), fiber cross-section (log-FC), and fiber density and cross-section (FDC) were computed [43]. A whole-brain tractogram of 20 million streamlines was generated from the template and filtered to 2 million streamlines using SIFT to reduce reconstruction biases [44]. Fixel-wise metrics were smoothed using connectivity-based smoothing (10-mm FWHM), which propagates information along structurally connected fixels. Statistical analysis was performed using the connectivity-based fixel enhancement (CFE) approach with 5000 permutations within a general linear model, comparing group means with age as a covariate [45]. No additional covariates were included in the model due to the heterogeneity of the patient cohort, which could reduce statistical power given the sample size.
Statistical analyses
Data normality was assessed with the Shapiro–Wilk test for each eye (right and left in controls; contralateral and ipsilateral to the most affected eye in patients). Depending on distribution, paired Student’s t-test or Wilcoxon signed-rank test was used to compare right vs left optic nerves in controls. As no significant differences were found, an average value per control subject was calculated. Average values of the optic nerves of the control group were then separately compared to metrics derived from the patients’ optic nerves according to their relation to the most affected eye using unpaired Welch’s t-test or Mann–Whitney test. Specifically, two independent comparisons were performed: control vs contralateral optic nerve and control vs ipsilateral optic nerve. Finally, paired comparisons were performed between the two optic nerves in patients. The same statistical approach was applied to the optic chiasm; however, the analysis was conducted at the fixel level. This statistical approach was applied to diffusion metrics derived from DTI, DKI, CSD, and MRDS. Fixel-based analyses (FBA) were conducted separately within a general linear model framework. Statistical significance was considered as p < 0.05. All statistical analyses were performed using GraphPad Prism (version 10, GraphPad Software, USA) and R (version 4.5.2).
Correlations between diffusion metrics and clinical parameters
Bivariate correlations were performed to explore the relationship between diffusion metrics and clinical parameters of glaucoma severity. Prior to the analysis, data normality was assessed using the Shapiro–Wilk test to determine whether Pearson or Spearman correlation coefficients should be applied. Pearson correlations were used when both variables followed a normal distribution, whereas Spearman correlations were used otherwise. Three patients were excluded from these analyses since their clinical parameters were not available, as they were evaluated in private practice and their ophthalmological records could not be accessed. These correlations are summarized in the matrix shown in Fig 5, where each cell with a dot indicates a significant correlation and the correlation coefficient (r) is color-coded. Additional correlations between diffusion metrics are provided in Supplementary S2 Fig. The statistical test used for each correlation is reported in the Supporting Information (S3 Table).
Results
In patients with asymmetric glaucoma, the nerve related to the most affected eye (ipsilateral) showed significantly altered values compared to the contralateral eye (Fig 2). Nearly all voxel-wise metrics showed clear asymmetry between the two optic nerves of patients, characterized by reduced FA, AFD, MK, RK, as well as increased MD, in the optic nerves ipsilateral to the most affected eye. The contralateral optic nerves of patients showed a statistically-significant difference to control subjects in FA and AFD.
Diffusion metrics (FA, MD, AFD, MK, AK, and RK) are shown separately for fibers stemming from right or left eyes in controls (blue), and from the contralateral (green) or most affected/ipsilateral eyes (red) in patients. Voxel-wise analysis was performed at the level of the optic nerves (A–F). Only statistically significant p-values are displayed.
Per-bundle analyses of the optic chiasm showed significant reductions of fixel-FA (MRDS) between the two fiber populations (i.e., stemming from ipsilateral and contralateral eyes), as well as between the affected eye and the control group, but not between the control group and the contralateral eye, consistent with findings from AFD using CSD (Fig 3). For fixel-MD, a significant increase was identified both intragroup (most affected vs contralateral eye) and intergroup (patients vs controls). Details about these comparisons are provided in the Supporting Information (S2 Table).
Fixel-wise metrics (fixel-FA, fixel-MD, and fixel-AFD) are shown separately for fibers stemming from right or left eyes in controls (blue), and from the most affected (ipsilateral, red) or contralateral (green) eyes in patients. Blue shading in the anatomical illustration indicates the analysis level at the optic chiasm. Only statistically significant p-values are displayed.
Fixel-based analysis revealed significant reductions in patients with asymmetric glaucoma in fiber density (FD), fiber cross-section (log-FC), and the combined metric FDC in both the optic tract and optic radiations compared to controls (Fig 4). These results reflect microstructural alterations (FD), macrostructural changes (log-FC), and combined effects (FDC). Only reductions of FD within the optic tracts and of FDC within both optic tracts and optic radiations were statistically significant after cluster-based correction for multiple comparisons.
A) Fiber density (FD), B) fiber cross-section (log-FC), and C) fiber density and cross-section (FDC) showed significant reductions in patients with asymmetric glaucoma (red) compared to healthy controls (blue). Top panels display statistical fixel maps overlaid on the population template. Fixels are visualized (black lines) if statistically significant (uncorrected p < 0.05), and color-coded (1-p) if significant after correction using connectivity-based fixel enhancement (CFE). Lower panels show group boxplots for each metric at the regions indicated by the arrows in the top panels.
Significant correlations were observed between diffusion MRI metrics and clinical parameters along the white matter of the visual pathway (Fig 5). As expected in an asymmetric glaucoma cohort, the ipsilateral eye showed stronger associations with clinical measures, such as FA correlating negatively with vCD and fixel-FA correlating positively with VF-MD. These examples illustrate the direct link between microstructural degeneration and clinical severity. In contrast, contralateral eyes exhibited a spectrum of damage, reflecting that in some patients the asymmetry was mainly due to differences in the degree of impairment between eyes. Consequently, correlations with clinical parameters were less consistent on the contralateral side; these results are presented in Supplementary S2 Fig. Additional correlations between diffusion metrics themselves are also provided in Supplementary S2 Fig and S3 Table from the Supporting Information.
Significant correlations are shown as circles (blue = negative, red = positive). Circle diameters reflect the significance of the correlation (p < 0.05), and crosses indicate trends (0.05 < p < 0.1). Scatter plots illustrate two exemplary associations. Clinical parameters and diffusion metrics of the optic nerve and chiasm correspond to the most affected eye; the average diffusion metrics of both hemispheres is used for the optic radiations. The gray shaded area around the regression lines represents the 95% confidence interval.
Discussion
The primary aim of this study was to evaluate the ability of clinically-feasible diffusion MRI methods to detect microstructural alterations in regions with complex fiber configurations. In this context, asymmetric glaucoma provides a biologically grounded model in which two fiber populations with different degrees of degeneration cross within the optic chiasm. This configuration offers a unique opportunity to assess the performance of advanced diffusion analyses in vivo. We show that bundle-wise analyses are capable of distinguishing the two axonal populations in the chiasm in vivo, in line with our previous reports in an animal model using ex vivo imaging [12]. Together, our findings provide support for the adequate interpretation of bundle-wise analyses throughout the rest of white matter. Furthermore, this study also provides evidence of white matter damage across the entire visual system in patients with asymmetric glaucoma. We show that different clinically-feasible diffusion metrics provide complementary information regarding the microarchitecture of the optic nerves, tracts, chiasm, and radiations. Moreover, these metrics correlate with disease severity, providing insights into the downstream degeneration of brain tissue secondary to glaucoma and, importantly, highlighting their potential as monitoring biomarkers.
The sensitivity of DTI for the detection of Wallerian degeneration in the optic nerve is well-known. Indeed, evaluation of the optic nerves of rodents after experimental retinal ischemia served as a test-bed for this purpose in the early days of DTI [46,47]. These and other evaluations identified the link between axonal injury with axial diffusivity, and axonal density and myelin abnormalities with radial diffusivity [2,3]. It follows that alterations of DTI metrics that indicate degeneration of the optic nerves have been demonstrated in patients with glaucoma [16–19]. Such findings are replicated in this work, yet, as compared with previous reports of patients with bilateral glaucoma, the population studied here allowed for the association of white matter damage to either the most affected or contralateral eyes. Importantly, the design of this study allows each patient to serve as their own internal control, thereby reducing the potential influence of systemic confounding factors such as diabetes or hypertension when comparing the most affected and contralateral pathways. However, the contralateral eye in asymmetric glaucoma is not necessarily unaffected, as early or subclinical degeneration may already be present. This is reflected in our results, where significant differences in FA and AFD were observed between the contralateral optic nerve of patients and healthy controls (Fig 2A,C). This should be considered when interpreting the observed asymmetries, which likely reflect differences in the extent of axonal damage rather than a comparison between intact and degenerated fiber populations. The severity of downstream degeneration of the optic nerves assessed with voxel-wise metrics derived from DTI and DKI matches the asymmetry of glaucoma (Fig 2). Furthermore, there is a tight correlation between diffusion metrics of the optic nerves and the degree of retinal alterations observed in OCT and visual field deviations (Fig 5). These results confirm the sensitivity of diffusion metrics to identify white matter degeneration in simple fiber configurations [48–50].
The optic nerves converge at the level of the optic chiasm, where around half of the axons (from the nasal hemisphere of each retina) cross the midline and interdigitate with axons stemming from the other eye. A major limitation of DTI is its inability to resolve crossing fibers, whereby fractional anisotropy is artifactually reduced in voxels with more than one fiber populations. This has hampered the application of DTI to study the optic chiasm in patients with glaucoma. The chiasm is a very discrete anatomical structure with a complex fiber configuration that can be inferred a priori. It therefore serves as an ideal region to provide evidence of the utility of advanced analyses of the diffusion signal. In this sense, the optic chiasm acts as a natural experimental model to evaluate whether diffusion methods can successfully disentangle and characterize fiber populations with different degrees of degeneration within the same voxel. This is not only relevant to the study of the optic pathway, but for the investigation of any region of the brain with crossing fibers, which can account for to up to two thirds of the human white matter [7]. Several methods have been proposed to resolve individual orientations of fiber populations that cross within a voxel [51], although not all possess the ability to derive per-bundle diffusion metrics [52]. In this work, we focus on two methods that reliably and independently characterize diffusion for each fiber population, with both methods having relatively lenient acquisition requirements that can be obtained with clinical scanners. The asymmetric presentation of glaucoma in the patients studied here provides a unique opportunity to assess the ability of these methods to detect per-bundle axonal alterations. We show that in the chiasm the multi-tensor fit performed with MRDS shows asymmetry of fixel-FA akin to that seen between the two optic nerves, where the fixels associated to the most affected eye had reduced fixel-FA (Fig 3A). Albeit to a lesser degree, there was also asymmetry of fixel-MD of the two fiber systems in the chiasm of patients, with a slight increase of the fibers related to the most affected eye (Fig 3C). Similar to the multi-tensor fit, CSD was also able to identify the most affected fiber population within the chiasm, which showed marked reductions of AFD (Fig 3C). However, in comparison to the marked asymmetry observed in the optic nerve, differences in the chiasm were less pronounced, possibly reflecting reduced sensitivity to detect subtle microstructural changes in regions with complex fiber configurations, and imperfect separation of the two fiber populations. In addition, fixel metrics derived from the multi-tensor fit and CSD (fixel-FA and fixel-AFD, respectively) were positively correlated with visual field deviations (Fig 5). These results in the chiasm of patients with asymmetric glaucoma are in line with our previous exploration of bundle-wise diffusion metrics in the chiasm of rodents with unilateral retinal ischemia, which directly and independently correlate with the number of axons in the corresponding optic nerves evaluated with quantitative histology [12]. In the human condition described here, the thickness of the RNFL and the vCD serve as indicators of retinal damage and act as proxies for the degree of axonal degeneration. However, only vCD showed a significant correlation with fixel-MD in the chiasm. The lack of correlations between RNFL and fixel-wise diffusion metrics is likely due to anatomical differences: while nearly all axons decussate at the level of the chiasm in rodents, only 50% cross the midline in humans [53]. The consistency between imaging metrics and clinical severity strengthens the interpretation of our findings and supports the validity of per-bundle diffusion metrics derived from clinically feasible acquisitions. Our current findings show that the ability to separate the diffusion signal into individual fiber bundles allows for a more precise characterization of microstructural changes in regions with crossing fibers. Importantly, these methods are applicable to other regions of the brain with crossing fibers and a wide range of neurodegenerative conditions, supporting the broader use of diffusion MRI to infer microstructural alterations in complex white-matter architectures.
White matter damage secondary to glaucoma is not restricted to the optic nerves, as trans-synaptic alterations have been reported in the optic radiations, and ultimately lead to reduced cortical thickness of the visual cortex (for a recent survey, see [54]). These findings indicate that, although originating in the eye, glaucoma is a condition with complex ramifications throughout the entire visual pathway. Whole-brain FBA clearly identified alterations of the optic radiations in patients with asymmetric glaucoma. Reductions of FD and FDC were observed bilaterally, as expected given the decussation of fibers in the optic chiasm. The three fixel-wise metrics of the optic radiation significantly correlated with vCD and VF-MD, demonstrating a tight link between clinical features and white matter abnormalities even in the most posterior aspects of the visual pathway.
Our findings are in agreement with previous reports that have shown alterations of diffusion metrics of the optic radiations in patients with glaucoma [55]. Voxel-wise analyses, most commonly based on DTI, DKI, or NODDI, have consistently reported reduced FA and increased MD and RD in the optic nerves, optic tracts, and optic radiations, often accompanied by decreases in axial diffusivity or kurtosis-based parameters [56–61]. These alterations correlate with clinical severity, including visual field mean deviation, vertical cup-to-disc ratio, and RNFL thickness. Notably, voxel-wise methods demonstrated that diffusion changes extend beyond the anterior visual pathway into the optic radiations and even extrastriate cortical regions. By contrast, the relatively fewer fixel-wise studies have provided bundle-specific evidence of degeneration. Using FBA, Haykal et al. reported significant reductions of FD, FC, and FDC in different segments of the optic nerve, showing correlations with clinical markers of glaucoma severity [59,62]. These findings highlight that fixel-wise metrics capture both microscopic loss of axonal density and macroscopic reductions in bundle cross-section, offering complementary sensitivity compared to voxel-wise indices.
Interestingly, the most direct assessment of retinal degeneration, namely RNFL, was correlated with diffusion metrics (FA and MK) in the optic nerves, but not in the rest of the visual pathway. For each patient we considered the average RNFL for each eye, which is an over-simplification of the retinal changes seen in glaucoma, that tend to be zonal rather than distributed. Although it is known that the entire visual pathway maintains a retinotopic distribution of fibers and one could theoretically match retinal quadrants with specific regions of white matter, the dMRI resolution achieved herein is insufficient to perform such a detailed analysis. The ability to distinguish two different fiber populations that cross within a voxel is proportional to the angle between them. With the current radial resolution of the diffusion signal used here, both MRDS and CSD should be able to disentangle two fiber populations separated by at least 35–45°. However, higher-resolution dMRI acquisitions have been shown to increase the detectability of fiber crossings and complex configurations [63]. Regardless, it is possible that the two methods may have identified only one fixel in an unknown portion of fixels within the chiasm, where in fact two fiber populations coexist. While conceptually and etymologically the chiasm resembles the greek letter , its morphology in humans can also resemble the letter H, with axons crossing the midline along the horizontal portion with antiparallel orientations. However, in coronal sections, the axons from either eye meet at an angle given their superior-inferior trajectories [64], thus improving their detection as two distinct fixels. An additional source of complexity stems from the anatomical organization of the human optic chiasm. Unlike rodents, in which nearly all axons decussate at the chiasm [65], primates exhibit a mixture of crossing and non-crossing axonal populations [66]. This arrangement yields an ambiguous “crossing versus kissing” fiber configuration at the chiasm [67]. Although our fixel selection strategy—aligning fixels with each optic nerve—was intended to identify the fiber populations arising from each eye, complete separation of crossing and non-crossing fibers cannot be ensured. Consequently, interpretations of ipsilateral versus contralateral differences should be approached with caution.
There are limitations to consider when interpreting these findings. First, the patient cohort was heterogeneous, and although clinical variables such as diabetes and hypertension were recorded, they were not included as covariates in the statistical analyses due to the limited sample size and the small and unbalanced distribution of these comorbidities between groups, which limits the robustness of covariate-based adjustments. Importantly, the design of this study, in which each patient serves as their own internal control, partially mitigates the potential confounding effects of such systemic conditions. Second, retinal degeneration was summarized using the average RNFL value for each eye, which may oversimplify the zonal nature of retinal damage due to glaucoma. Third, the optic chiasm is a very small structure, and although the resolution used here is higher than typical dMRI acquired in clinical settings, partial volume effects are non-negligible. Future studies with higher-resolution diffusion imaging and larger cohorts may allow more detailed characterization of these microstructural changes.
Conclusion
Diffusion metrics identified white matter abnormalities throughout the visual pathway of patients with glaucoma indicative of Wallerian and trans-synaptic degeneration, and are useful imaging biomarkers for severity of the disease. Our results provide evidence of the ability of advanced dMRI methods to robustly evaluate white matter characteristics in white matter, even in regions with fiber crossings, where each fiber population can be assessed independently. By extension, our results allow for better interpretation of fixel-wise analyses of white matter beyond the optic pathway.
Supporting information
S1 Fig. Comparison of MUSE and Hyperband (HB) acquisitions.
Raw and preprocessed images and derived maps are shown. The first two columns show the average images at b = 0 and b = 2500 s/mm², respectively, followed by FA and principal diffusivity direction color-coded maps (V1) with boxes indicating the magnified regions in the rightmost column. Even before preprocessing, the MUSE acquisition shows overall better image quality and improved maps of diffusion metrics. The region of fiber decussation at the level of chiasm is better observed in the MUSE acquisition as a red region indicating fibers crossing the midline (long arrow in the enlarged V1 map). Contrarily, the HB acquisition shows considerable signal pile-up (yellow arrowheads) that result in inadequate estimation of diffusion metrics within the region that corresponds to the optic chiasm and tracts (white arrowheads in enlarged V1). While these artifacts are minimized after preprocessing the HB acquisition, the decussation of fibers is mostly missing.
https://doi.org/10.1371/journal.pone.0349951.s001
(PDF)
S2 Fig. Correlation matrix of all diffusion metrics and clinical parameters organized ipsilateral and contralateral to the most affected eyes.
Significant correlations are shown as circles (blue = negative, red = positive). Circle diameters reflect the significance of the correlation (p < 0.05), and crosses indicate trends (0.05 < p < 0.1). The average diffusion metrics of both hemispheres is used for the optic radiations.
https://doi.org/10.1371/journal.pone.0349951.s002
(PDF)
Acknowledgments
We thank Dr. Erick Pasaye for support during MRI acquisition. Data analysis was partially performed at the National Laboratory for advanced scientific visualization (LAVIS), with help from Luis Aguilar. We also appreciate the technical support provided by Mirelta Regalado, Leopoldo González-Santos, Juan Ortiz-Retana and Moisés Baltazar. Dr. Ricardo Ríos and Dr. Stéphanie Thebault provided useful ideas and suggestions during the course of this project.
References
- 1. Moseley ME, Cohen Y, Kucharczyk J, Mintorovitch J, Asgari HS, Wendland MF, et al. Diffusion-weighted MR imaging of anisotropic water diffusion in cat central nervous system. Radiology. 1990;176(2):439–45. pmid:2367658
- 2. Beaulieu C. The basis of anisotropic water diffusion in the nervous system - a technical review. NMR Biomed. 2002;15(7–8):435–55. pmid:12489094
- 3. Concha L. A macroscopic view of microstructure: using diffusion-weighted images to infer damage, repair, and plasticity of white matter. Neuroscience. 2014;276:14–28. pmid:24051366
- 4. Basser PJ, Mattiello J, Le Bihan D. MR diffusion tensor spectroscopy and imaging. Biophysical Journal. 1994;66(1):259–67.
- 5. Behrens TEJ, Berg HJ, Jbabdi S, Rushworth MFS, Woolrich MW. Probabilistic diffusion tractography with multiple fibre orientations: What can we gain?. Neuroimage. 2007;34(1):144–55. pmid:17070705
- 6. Tournier J-D, Mori S, Leemans A. Diffusion tensor imaging and beyond. Magn Reson Med. 2011;65(6):1532–56. pmid:21469191
- 7. Jeurissen B, Leemans A, Tournier J-D, Jones DK, Sijbers J. Investigating the prevalence of complex fiber configurations in white matter tissue with diffusion magnetic resonance imaging. Hum Brain Mapp. 2013;34(11):2747–66. pmid:22611035
- 8. Douaud G, Jbabdi S, Behrens TEJ, Menke RA, Gass A, Monsch AU, et al. DTI measures in crossing-fibre areas: increased diffusion anisotropy reveals early white matter alteration in MCI and mild Alzheimer’s disease. Neuroimage. 2011;55(3):880–90. pmid:21182970
- 9. Coronado-Leija R, Ramirez-Manzanares A, Marroquin JL. Estimation of individual axon bundle properties by a Multi-Resolution Discrete-Search method. Med Image Anal. 2017;42:26–43. pmid:28735215
- 10. Tournier J-D, Calamante F, Connelly A. Robust determination of the fibre orientation distribution in diffusion MRI: non-negativity constrained super-resolved spherical deconvolution. Neuroimage. 2007;35(4):1459–72. pmid:17379540
- 11. Raffelt D, Tournier J-D, Rose S, Ridgway GR, Henderson R, Crozier S, et al. Apparent Fibre Density: a novel measure for the analysis of diffusion-weighted magnetic resonance images. Neuroimage. 2012;59(4):3976–94. pmid:22036682
- 12. Rojas-Vite G, Coronado-Leija R, Narvaez-Delgado O, Ramírez-Manzanares A, Marroquín JL, Noguez-Imm R, et al. Histological validation of per-bundle water diffusion metrics within a region of fiber crossing following axonal degeneration. Neuroimage. 2019;201:116013. pmid:31326575
- 13. Hernandez-Gutierrez E, Coronado-Leija R, Edde M, Dumont M, Houde J-C, Barakovic M, et al. Multi-tensor fixel-based metrics in tractometry: application to multiple sclerosis. Front Neurosci. 2024;18:1467786. pmid:39758886
- 14. Tham Y-C, Li X, Wong TY, Quigley HA, Aung T, Cheng C-Y. Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis. Ophthalmology. 2014;121(11):2081–90. pmid:24974815
- 15. Weinreb RN, Aung T, Medeiros FA. The pathophysiology and treatment of glaucoma: a review. JAMA. 2014;311(18):1901–11. pmid:24825645
- 16. Garaci FG, Bolacchi F, Cerulli A, Melis M, Spanò A, Cedrone C, et al. Optic nerve and optic radiation neurodegeneration in patients with glaucoma: in vivo analysis with 3-T diffusion-tensor MR imaging. Radiology. 2009;252(2):496–501. pmid:19435941
- 17. Li K, Lu C, Huang Y, Yuan L, Zeng D, Wu K. Alteration of fractional anisotropy and mean diffusivity in glaucoma: novel results of a meta-analysis of diffusion tensor imaging studies. PLoS One. 2014;9(5):e97445. pmid:24828063
- 18. Mendoza M, Shotbolt M, Faiq MA, Parra C, Chan KC. Advanced Diffusion MRI of the Visual System in Glaucoma: From Experimental Animal Models to Humans. Biology (Basel). 2022;11(3):454. pmid:35336827
- 19. Cooper AC, Tchernykh M, Shmuel A, Mendola JD. Diffusion tensor imaging of optic neuropathies: a narrative review. Quant Imaging Med Surg. 2024;14(1):1086–107. pmid:38223128
- 20. Miki A, Okazaki T, Weinreb RN, Morota M, Tanimura A, Kawashima R, et al. Evaluating Visual Field Progression in Advanced Glaucoma Using Trend Analysis of Targeted Mean Total Deviation. J Glaucoma. 2022;31(4):235–41. pmid:35019876
- 21. Leung CK, Cheung CYL, Weinreb RN, Qiu K, Liu S, Li H, et al. Evaluation of retinal nerve fiber layer progression in glaucoma: a study on optical coherence tomography guided progression analysis. Invest Ophthalmol Vis Sci. 2010;51(1):217–22. pmid:19684001
- 22. Chen N-K, Guidon A, Chang H-C, Song AW. A robust multi-shot scan strategy for high-resolution diffusion weighted MRI enabled by multiplexed sensitivity-encoding (MUSE). Neuroimage. 2013;72:41–7. pmid:23370063
- 23. Jeong H-K, Dewey BE, Hirtle JAT, Lavin P, Sriram S, Pawate S, et al. Improved diffusion tensor imaging of the optic nerve using multishot two-dimensional navigated acquisitions. Magn Reson Med. 2015;74(4):953–63. pmid:25263603
- 24. Takemura H, Liu W, Kuribayashi H, Miyata T, Kida I. Evaluation of simultaneous multi-slice readout-segmented diffusion-weighted MRI acquisition in human optic nerve measurements. Magn Reson Imaging. 2023;102:103–14. pmid:37149064
- 25. Veraart J, Novikov DS, Christiaens D, Ades-Aron B, Sijbers J, Fieremans E. Denoising of diffusion MRI using random matrix theory. Neuroimage. 2016;142:394–406. pmid:27523449
- 26. Cordero-Grande L, Christiaens D, Hutter J, Price AN, Hajnal JV. Complex diffusion-weighted image estimation via matrix recovery under general noise models. Neuroimage. 2019;200:391–404. pmid:31226495
- 27.
Fadnavis S, Batson J, Garyfallidis E. Patch2Self: Denoising Diffusion MRI with Self-Supervised Learning. 2020.
- 28. Kellner E, Dhital B, Kiselev VG, Reisert M. Gibbs-ringing artifact removal based on local subvoxel-shifts. Magn Reson Med. 2016;76(5):1574–81. pmid:26745823
- 29. Andersson JLR, Sotiropoulos SN. An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging. Neuroimage. 2016;125:1063–78. pmid:26481672
- 30. Tustison NJ, Avants BB, Cook PA, Zheng Y, Egan A, Yushkevich PA, et al. N4ITK: improved N3 bias correction. IEEE Trans Med Imaging. 2010;29(6):1310–20. pmid:20378467
- 31. Garyfallidis E, Brett M, Amirbekian B, Rokem A, van der Walt S, Descoteaux M. Dipy, a library for the analysis of diffusion MRI data. Front Neuroinform. 2014;8:8.
- 32. Henriques RN, Correia MM, Marrale M, Huber E, Kruper J, Koudoro S, et al. Diffusional Kurtosis Imaging in the Diffusion Imaging in Python Project. Front Hum Neurosci. 2021;15:675433. pmid:34349631
- 33. Jensen JH, Helpern JA, Ramani A, Lu H, Kaczynski K. Diffusional kurtosis imaging: The quantification of non‐gaussian water diffusion by means of magnetic resonance imaging. Magn Reson Med. 2005;53(6):1432–40.
- 34. Tournier J-D, Smith R, Raffelt D, Tabbara R, Dhollander T, Pietsch M, et al. MRtrix3: A fast, flexible and open software framework for medical image processing and visualisation. Neuroimage. 2019;202:116137. pmid:31473352
- 35. Tax CMW, Jeurissen B, Vos SB, Viergever MA, Leemans A. Recursive calibration of the fiber response function for spherical deconvolution of diffusion MRI data. Neuroimage. 2014;86:67–80. pmid:23927905
- 36.
Karaman M, Mito R, Powell E, Rheault F, Winzeck S. Computational Diffusion MRI: 14th International Workshop, CDMRI 2023, Held in Conjunction with MICCAI 2023, Vancouver, BC, Canada, October 8, 2023, Proceedings. In: Lecture Notes in Computer Science. Cham: Springer Nature Switzerland; 2023.
- 37. Dhollander T, Emsell L, Van Hecke W, Maes F, Sunaert S, Suetens P. Track orientation density imaging (TODI) and track orientation distribution (TOD) based tractography. NeuroImage. 2014;94:312–36.
- 38.
Hernandez-Gutierrez E, Coronado-Leija R, Ramirez-Manzanares A, Barakovic M, Magon S, Descoteaux M. Improving Multi-Tensor Fitting with Global Information from Track Orientation Density Imaging. In: Karaman M, Mito R, Powell E, Rheault F, Winzeck S, editors. Computational Diffusion MRI. Cham: Springer Nature Switzerland; 2023. p. 35–46.
- 39.
Dhollander T, Raffelt D, Connelly A. Unsupervised 3-tissue response function estimation from single-shell or multi-shell diffusion MR data without a co-registered T1 image. In: ISMRM workshop on breaking the barriers of diffusion MRI. vol. 5. Lisbon; 2016. p. 1.
- 40.
Dhollander T, Mito R, Raffelt D, Connelly A. Improved white matter response function estimation for 3-tissue constrained spherical deconvolution. In: Proc Intl Soc Mag Reson Med. vol. 555. 2019. p. 107–20.
- 41. Jeurissen B, Tournier J-D, Dhollander T, Connelly A, Sijbers J. Multi-tissue constrained spherical deconvolution for improved analysis of multi-shell diffusion MRI data. Neuroimage. 2014;103:411–26. pmid:25109526
- 42. Raffelt D, Tournier J-D, Fripp J, Crozier S, Connelly A, Salvado O. Symmetric diffeomorphic registration of fibre orientation distributions. Neuroimage. 2011;56(3):1171–80. pmid:21316463
- 43. Raffelt DA, Tournier J-D, Smith RE, Vaughan DN, Jackson G, Ridgway GR, et al. Investigating white matter fibre density and morphology using fixel-based analysis. Neuroimage. 2017;144(Pt A):58–73. pmid:27639350
- 44. Smith RE, Tournier J-D, Calamante F, Connelly A. SIFT: Spherical-deconvolution informed filtering of tractograms. Neuroimage. 2013;67:298–312. pmid:23238430
- 45. Raffelt DA, Smith RE, Ridgway GR, Tournier J-D, Vaughan DN, Rose S, et al. Connectivity-based fixel enhancement: Whole-brain statistical analysis of diffusion MRI measures in the presence of crossing fibres. Neuroimage. 2015;117:40–55. pmid:26004503
- 46. Song S-K, Sun S-W, Ju W-K, Lin S-J, Cross AH, Neufeld AH. Diffusion tensor imaging detects and differentiates axon and myelin degeneration in mouse optic nerve after retinal ischemia. Neuroimage. 2003;20(3):1714–22. pmid:14642481
- 47. Sun S-W, Liang H-F, Cross AH, Song S-K. Evolving Wallerian degeneration after transient retinal ischemia in mice characterized by diffusion tensor imaging. Neuroimage. 2008;40(1):1–10. pmid:18187343
- 48. Zhu Y, Wang R, Pappas AC, Seifert P, Savol A, Sadreyev RI, et al. Astrocytes in the Optic Nerve Are Heterogeneous in Their Reactivity to Glaucomatous Injury. Cells. 2023;12(17):2131. pmid:37681863
- 49. García-Bermúdez MY, Freude KK, Mouhammad ZA, van Wijngaarden P, Martin KK, Kolko M. Glial Cells in Glaucoma: Friends, Foes, and Potential Therapeutic Targets. Front Neurol. 2021;12:624983. pmid:33796062
- 50. Salkar A, Palanivel V, Basavarajappa D, Mirzaei M, Schulz A, Yan P, et al. Glial and immune dysregulation in glaucoma independent of retinal ganglion cell loss: a human post-mortem histopathology study. Acta Neuropathol Commun. 2025;13(1):141. pmid:40581633
- 51. Caan MWA, Khedoe HG, Poot DHJ, den Dekker AJ, Olabarriaga SD, Grimbergen KA, et al. Estimation of diffusion properties in crossing fiber bundles. IEEE Trans Med Imaging. 2010;29(8):1504–15. pmid:20562045
- 52. Mishra V, Guo X, Delgado MR, Huang H. Toward tract-specific fractional anisotropy (TSFA) at crossing-fiber regions with clinical diffusion MRI. Magn Reson Med. 2015;74(6):1768–79.
- 53. Jeffery G, Erskine L. Variations in the architecture and development of the vertebrate optic chiasm. Prog Retin Eye Res. 2005;24(6):721–53. pmid:16027026
- 54. Sharma S, Chitranshi N, Wall RV, Basavarajappa D, Gupta V, Mirzaei M, et al. Trans-synaptic degeneration in the visual pathway: Neural connectivity, pathophysiology, and clinical implications in neurodegenerative disorders. Surv Ophthalmol. 2022;67(2):411–26. pmid:34146577
- 55. Kasa LW, Donovan S, Kwon E, Holdsworth S, Schierding W, Danesh-Meyer H. Application of advanced diffusion MRI based tractometry of the visual pathway in glaucoma: a systematic review. Front Neurosci. 2025;19:1577991. pmid:40470296
- 56. Tellouck L, Durieux M, Coupé P, Cougnard-Grégoire A, Tellouck J, Tourdias T, et al. Optic Radiations Microstructural Changes in Glaucoma and Association With Severity: A Study Using 3Tesla-Magnetic Resonance Diffusion Tensor Imaging. Invest Ophthalmol Vis Sci. 2016;57(15):6539–47. pmid:27918827
- 57. Zhou W, Muir ER, Chalfin S, Nagi KS, Duong TQ. MRI Study of the Posterior Visual Pathways in Primary Open Angle Glaucoma. J Glaucoma. 2017;26(2):173–81. pmid:27661989
- 58. Miller N, Liu Y, Krivochenitser R, Rokers B. Linking neural and clinical measures of glaucoma with diffusion magnetic resonance imaging (dMRI). PLoS One. 2019;14(5):e0217011. pmid:31150402
- 59. Haykal S, Jansonius NM, Cornelissen FW. Investigating changes in axonal density and morphology of glaucomatous optic nerves using fixel-based analysis. Eur J Radiol. 2020;133:109356. pmid:33129102
- 60. Ogawa S, Takemura H, Horiguchi H, Miyazaki A, Matsumoto K, Masuda Y, et al. Multi-Contrast Magnetic Resonance Imaging of Visual White Matter Pathways in Patients With Glaucoma. Invest Ophthalmol Vis Sci. 2022;63(2):29. pmid:35201263
- 61. Kruper J, Richie-Halford A, Benson NC, Caffarra S, Owen J, Wu Y, et al. Convolutional neural network-based classification of glaucoma using optic radiation tissue properties. Commun Med (Lond). 2024;4(1):72. pmid:38605245
- 62. Haykal S, Curcic-Blake B, Jansonius NM, Cornelissen FW. Fixel-Based Analysis of Visual Pathway White Matter in Primary Open-Angle Glaucoma. Invest Ophthalmol Vis Sci. 2019;60(12):3803–12. pmid:31504081
- 63. Schilling K, Gao Y, Janve V, Stepniewska I, Landman BA, Anderson AW. Can increased spatial resolution solve the crossing fiber problem for diffusion MRI? NMR Biomed. 2017;30(12):10.1002/nbm.3787.
- 64. Sarlls JE, Pierpaoli C. In vivo diffusion tensor imaging of the human optic chiasm at sub-millimeter resolution. Neuroimage. 2009;47(4):1244–51. pmid:19520170
- 65. Jeffery G, Erskine L. Variations in the architecture and development of the vertebrate optic chiasm. Prog Retin Eye Res. 2005;24(6):721–53. pmid:16027026
- 66. He J, Zhang F, Xie G, Yao S, Feng Y, Bastos DCA, et al. Comparison of multiple tractography methods for reconstruction of the retinogeniculate visual pathway using diffusion MRI. Hum Brain Mapp. 2021;42(12):3887–904. pmid:33978265
- 67. Roebroeck A, Galuske R, Formisano E, Chiry O, Bratzke H, Ronen I. High-Resolution Diffusion Tensor Imaging and Tractography of the Human Optic Chiasm at 9.4 T. NeuroImage. 2008;39(1):157–68.