Biology subjects

Rokem, A.

Publications and source records attributed to Rokem, A..

At least 19 recordsLinked to original sources

bioRxiv2026

A Comparative Evaluation of Structural MRI Foundation Models for Brain Age Regression and Sex Classification

Foundation models pre-trained on large neuroimaging datasets offer a promising approach to overcome the limited sample sizes typical of clinical imaging studies, yet their generalization across diverse populations remains unclear. We present the first systematic benchmark of four publicly available structural MRI foundation models: AnatCL, BrainIAC, 3D-Neuro-SimCLR, and SwinBrain. Using T1-weighted MRIs from the Parkinson's Progression Markers Initiative (PPMI), Healthy Brain Network (HBN), and Nathan Kline Institute (NKI) datasets, we evaluate these models on sex classification, brain age prediction, and body mass index prediction, comparing against models trained from FreeSurfer-derived cortical thickness and cortical surface area features. Submitted models are evaluated using a standardized frozen feature probing framework. The evaluation methods are available in BrainFMBench, a living benchmark for structural brain MRI foundation models hosted on GitHub, where new models can be added through pull requests. Although some foundation models outperformed FreeSurfer on particular tasks and datasets, 3D-Neuro-SimCLR and AnatCL outperformed the baselines overall, with 3D-Neuro-SimCLR demonstrating the most consistent performance (with the notable exception of HBN sex classification). The remaining models did not consistently outperform the baselines, indicating that the advantage of learned representations over morphometric features was not consistent across datasets and tasks for these models. In addition, cross-model feature correlation analysis reveals that foundation model representations correlate differently with traditional cortical measurements. These findings position structural MRI foundation models, particularly 3D-Neuro-SimCLR and AnatCL, as promising avenues to boost the performance of predictive models in neuroimaging.

neuroscience↗

bioRxiv2026

Neonatal brain-age models in full- and preterm infants

Prematurity affects brain development and increases risk for neurodevelopmental impairments. Yet reliable biomarkers for at-risk infants remain limited. The goals of this study are (1a) to construct a white matter neonatal brain-age model including full-term and preterm neonates from a large publicly-available data set, (1b) to evaluate the accuracy of this model for characterizing the preterm brain from the same data set; (2) to determine if a similar model can predict brain-age based on clinical MRI scans from high-risk neonates born preterm; and (3) to evaluate whether this predictive model provides information about the infants health beyond conventional clinical and demographic measures. We developed brain-age prediction models using diffusion magnetic resonance imaging-derived white matter features from two datasets: (1) the developing Human Connectome Project (dHCP; 368 healthy infants) and (2) a clinical sample collected at the Lucile Packard Childrens Hospital (LPCH; 162 high-risk preterm infants). White matter features demonstrated strong predictive performance in the dHCP dataset (within 3.9 days) and the LPCH clinical dataset (within 6.6 days). However, brain-age metrics (i.e., brain-age gap) showed no significant associations with health complications measured by a composite score of common prematurity complications. While tractometry-derived brain-age models accurately characterize brain maturation in the neonatal brain, their sensitivity to clinical complications in preterm infants appears limited. Global white matter maturation measures derived from clinical grade data may be insufficiently sensitive to capture the cumulative burden of prematurity-related morbidities, suggesting need for multimodal or longitudinal biomarkers. HIGHLIGHTSBrain-age models of neonatal dMRI data predict age within days. Brain-age models of preterm infants predict post-menstrual age within a week. Brain-age models are accurate with research and clinical dMRI data in preterm infants. Brain-age gap was not sensitive to health status in a clinical preterm sample. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=79 SRC="FIGDIR/small/701337v1_ufig1.gif" ALT="Figure 1000"> View larger version (15K): org.highwire.dtl.DTLVardef@16f5c5dorg.highwire.dtl.DTLVardef@1918e0org.highwire.dtl.DTLVardef@1b0246eorg.highwire.dtl.DTLVardef@127f132_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2025

Cortical and white matter T1w/T2w development proceed in concert during early infancy

The infant brain undergoes rapid myelination that is critical for healthy brain function. This development has been characterized for gray and white matter independently, but the link between gray and white matter myelination remains unexplored. To close this knowledge gap, we evaluated two complementary myelin-sensitive imaging metrics: Large-scale (N=273) T1w/T2w and quantitative (N=21) R1 data. Automated software was employed to identify 26 white matter bundles and map their cortical terminations, before evaluating T1w/T2w and R1 development shortly after birth. Here we show that for both metrics mean values as well as developmental slopes are correlated across tissues. The synchrony of brain T1w/T2w is impacted by gestational age and prematurity, whereas inter-individual differences in this synchrony predict motor outcomes at 17 - 25 months of age. As T1w/T2w and R1 are associated with myelin content, our results reveal an intricate relationship between gray and white matter myelination.

neuroscience↗

bioRxiv2025

Vessels hiding in plain sight: quantifying brain vascular morphology in anatomical MR images using deep learning

Non-invasive assessment of brain blood vessels with magnetic resonance (MR) imaging provides important information about brain health and aging. Time-of-flight MR angiography (TOF-MRA) in particular is commonly used to assess the morphology of blood vessels, but acquisition of MRA is time-consuming and is not as commonly employed in research studies or in the clinic as the more standard T1- or T2-weighted MR contrasts (T1w/T2w). To enable quantification of brain blood vessel morphology in T1w/T2w images, we trained a neural network model, anat2vessels, on a dataset with paired MR/MRA. The model provides accurate segmentations as assessed in cross-validation on ground truth images, particularly in cases where T2w images are used. In addition, correlation between features that are extracted from model-based vessel segmentations and from ground truth account for as much as 78% of the variance in these features. We further evaluated the model in another dataset that does not include MRA and found that anat2vessels-based vessel morphology features contain information about aging that is not captured by cortical thickness features that are routinely extracted from T1w/T2w images. Moreover, we found that vessel morphology features are associated with individual variability in blood pressure and cognitive abilities. Taken together these results suggest that anat2vessels could be fruitfully applied to a range of existing and new datasets to assess the role of brain blood vessels in aging and brain health. The methods are provided as open-source software in https://github.com/nrdg/anat2vessels/.

neuroscience↗

bioRxiv2025

Two Axes of White Matter Development

Despite decades of neuroimaging research, how white matter develops along the length of major tracts in humans remains unknown. Here, we identify fundamental patterns of white matter maturation by examining developmental variation along major, long-range cortico-cortical tracts in youth ages 5-23 years using diffusion MRI from three large-scale, cross-sectional datasets (total N = 2,716). Across datasets, we delineate two replicable axes of human white matter development. First, we find a deep-to-superficial axis, in which superficial tract regions near the cortical surface exhibit greater age-related change than deep tract regions. Second, we demonstrate that the development of superficial tract regions aligns with the cortical hierarchy defined by the sensorimotor-association axis, with tract ends adjacent to sensorimotor cortices maturing earlier than those adjacent to association cortices. These results reveal developmental variation along tracts that conventional tract-average analyses have previously obscured, challenging the implicit assumption that white matter tracts mature uniformly along their length. Such developmental variation along tracts may have functional implications, including mitigating ephaptic coupling in densely packed deep tract regions and tuning neural synchrony through hierarchical development in superficial tract regions - ultimately refining neural transmission in youth.

neuroscience↗

bioRxiv2025

Reproducible Brain Charts: An open data resource for mapping brain development and its associations with mental health

Major mental disorders are increasingly understood as disorders of brain development. Large and heterogeneous samples are required to define generalizable links between brain development and psychopathology. To this end, we introduce the Reproducible Brain Charts (RBC), an open data resource that integrates data from 5 large studies of brain development in youth from three continents (N=6,346; 45% Female). Confirmatory bifactor models were used to create harmonized psychiatric phenotypes that capture major dimensions of psychopathology. Following rigorous quality assurance, neuroimaging data were carefully curated and processed using consistent pipelines in a reproducible manner with DataLad, the Configurable Pipeline for the Analysis of Connectomes (C-PAC), and FreeSurfer. Initial analyses of RBC data emphasize the benefit of careful quality assurance and data harmonization in delineating developmental effects and associations with psychopathology. Critically, all RBC data - including harmonized psychiatric phenotypes, unprocessed images, and fully processed imaging derivatives - are openly shared without a data use agreement via the International Neuroimaging Data-sharing Initiative. Together, RBC facilitates large-scale, reproducible, and generalizable research in developmental and psychiatric neuroscience.

neuroscience↗

bioRxiv2024

Free water elimination tractometry for aging brains

Tractometry of diffusion-weighted magnetic resonance imaging (dMRI) non-invasively quantifies tissue properties of brain connections. It is widely used in aging studies but could be less reliable in aging brains due to increased white matter free water. We demonstrate that computational free water elimination (FWE) increases reliability and accuracy of tractometry in a large (n = 339) cohort of older adults (66 - 103 y.o.). We found substantial (up to [~]37%) improvements in reliability in a split-half comparison at every stage of the pipeline: estimation of voxel-level fiber orientation distribution functions, delineation of major pathway trajectories, and assessment of tissue properties along the pathways. FWE also improves inferences from tractometry, producing more accurate cross-validated predictions of clinician Fazekas scores. By sub-sampling a multi-b-value dataset, we demonstrated that these findings generalize to both single-b-value data, which is important for many datasets where only one b-value may be available. Overall, the results highlight the importance of accounting for free water in tractometry studies, especially in aging brains. We provide opensource software for free-water elimination that can be applied to a wide range of clinical and research datasets (https://github.com/nrdg/fwe).

neuroscience↗

bioRxiv2024

Development of the Left Arcuate Fasciculus is Linked to Learning Gains in Reading, but not Math

Past studies leveraging cross-sectional data have raised questions surrounding the relationship between diffusion properties of the white matter and academic skills. Some studies have suggested that white matter properties serve as static predictors of academic skills, whereas other studies have observed no such relationship. On the other hand, longitudinal studies have suggested that within-individual changes in the white matter are linked to learning gains over time. In the present study, we look to replicate and extend the previous longitudinal results linking longitudinal changes in the white matter properties of the left arcuate fasciculus to individual differences in reading development. To do so, we analyzed diffusion MRI data, along with reading and mathematics scores in a longitudinal sample of 340 students as they progressed from 1st grade into 4th grade. Longitudinal growth models revealed that year-to-year within-individual changes in reading scores, but not math, were related to the development of the left arcuate fasciculus. These findings provide further evidence linking the dynamics of white matter development and learning in a unique sample and highlight the importance of longitudinal designs.

neuroscience↗

bioRxiv2024

Assessing white matter plasticity in a randomized controlled trial of early literacy training in preschoolers

Reading is a cognitive skill that requires our brain to go through a myriad of changes during learning. While many studies have described how reading acquisition shapes childrens brain function, less is known about the impact of reading on brain structure. Here we examined short-term causal effects of reading training on preschoolers behavior and white matter structure. Forty-eight English-speaking preschoolers (4y10m to 6y2m) participated in a randomized controlled trial where they were randomly assigned to two training programs: the Letter training program was focused on key skills for reading (e.g., decoding and letter knowledge), while the Language training program strengthened oral language comprehension skills without exposure to text. Longitudinal behavioral data showed that only the Letter Training group increased letter knowledge and decoding skills after the two-week training. Diffusion MRI measures (FA and MD) of eighteen white matter pathways (including the left arcuate and the left inferior longitudinal fasciculus) did not reveal any statistically significant changes for either group despite high degrees of scan-rescan reliability across sessions. These findings suggest that a two- week reading training program can cause changes in preschoolers letter knowledge and decoding abilities, without being accompanied by measurable changes in the diffusion properties of the major white matter pathways of the reading network. We conclude highlighting possible constraints (i.e., age, training onset and duration, cognitive profile) to reading-related white matter plasticity.

neuroscience↗

bioRxiv2023

Educational Environment and White Matter Development in Early Adolescence

Coarse measures of socioeconomic status, such as parental income or parental education, have been linked to differences in white matter development. However, these measures do not provide insight into specific aspects of an individuals environment and how they relate to brain development. On the other hand, educational intervention studies have shown that changes in an individuals educational context can drive measurable changes in their white matter. These studies, however, rarely consider socioeconomic factors in their results. In the present study, we examined the unique effect of educational opportunity on white matter development, even when controlling other known socioeconomic factors. To explore this question, we leveraged the rich demographic and neuroimaging data available in the ABCD study, as well the unique data-crosswalk between ABCD and the Stanford Education Data Archive (SEDA). We find that educational opportunity is related to accelerated white matter development, even when accounting for other socioeconomic factors, and that this relationship is most pronounced in white matter tracts associated with academic skills. These results suggest that the school a child attends has a measurable impact on brain development for years to come.

neuroscience↗

bioRxiv2023

Non-Stationary Dynamic Mode Decomposition

Many physical processes display complex high-dimensional time-varying behavior, from global weather patterns to brain activity. An outstanding challenge is to express high dimensional data in terms of a dynamical model that reveals their spatiotemporal structure. Dynamic Mode Decomposition is a means to achieve this goal, allowing the identification of key spatiotemporal modes through the diagonalization of a finite dimensional approximation of the Koopman operator. However, DMD methods apply best to time-translationally invariant or stationary data, while in many typical cases, dynamics vary across time and conditions. To capture this temporal evolution, we developed a method, Non-Stationary Dynamic Mode Decomposition (NS-DMD), that generalizes DMD by fitting global modulations of drifting spatiotemporal modes. This method accurately predicts the temporal evolution of modes in simulations and recovers previously known results from simpler methods. To demonstrate its properties, the method is applied to multi-channel recordings from an awake behaving non-human primate performing a cognitive task.

neuroscience↗

bioRxiv2023

Incremental improvements in tractometry-based brain-age modeling with deep learning

Multivariate measurements of human brain white matter (WM) with diffusion MRI (dMRI) provide information about the role of WM in a variety of cognitive functions and in brain health. Statistical models take advantage of the regularities in these data to make inferences about individual differences. For example, dMRI data provide the basis for accurate brain-age models - models that predict the chronological age of participants from WM tissue properties. Deep learning (DL) models are powerful machine learning models, which have been shown to provide benefits in many multivariate analysis settings. We investigated whether DL would provide substantial improvements for brain-age models based on dMRI measurements of WM in a large sample of children and adolescents. We found that some DL models fit the data better than a linear baseline, but the differences are small. In particular, recurrent neural network architectures provide up to [~]6% improvement in accuracy. This suggests that information about WM development is mostly accessible with linear models, and does not require the additional invariance and non-linearity offered by DL models. However, in some applications this incremental improvement may prove critical. We provide open-source software that fits DL models to dMRI data (https://yeatmanlab.github.io/AFQ-Insight).

neuroscience↗

bioRxiv2023

Human white matter myelination rate slows down at birth

The formation of myelin, the fatty sheath that insulates nerve fibers, is critical for healthy brain function. A fundamental open question is what is the impact of being born on myelin growth. To address this question, we evaluated a large (n=300) cross-sectional sample of newborns from the Developing Human Connectome Project (dHCP). First, we developed new software for the automated identification of 20 white matter bundles in individuals that is well-suited for large samples. Next, we fit linear models that quantify T1w/T2w, a myelin-sensitive imaging contrast, increases along bundles. We found faster growth of T1w/T2w along the lengths of all bundles before birth than right after birth. Further, in a separate longitudinal sample of preterm infants (N=34), we found lower T1w/T2w at term-equivalent age than in full-term peers. By applying the linear models fit on the cross-section sample to the longitudinal sample of preterm infants, we find that their delay in T1w/T2w growth is well explained by the amount of time preterm infants spend developing in utero and ex utero. These results suggest that being born slows the rate of myelin growths. This reduction in the rate of myelin growth at birth, in turn, explains lower myelin content in individuals born preterm, and could account for long-term cognitive, neurological, and developmental consequences of preterm birth. We hypothesize that closely matching the environment of infants born preterm to what they would have experienced in the womb may reduce delays in myelin growth and hence improve developmental outcomes.

neuroscience↗

bioRxiv2023

Specific and non-linear effects of glaucoma on optic radiation tissue properties

Changes in sensory input with aging and disease affect brain tissue properties. To establish the link between glaucoma, the most prevalent cause of irreversible blindness, and changes in major brain connections, we characterized white matter tissue properties in diffusion MRI measurements in a large sample of subjects with glaucoma (N=905; age 49-80) and healthy controls (N=5,292; age 45-80) from the UK Biobank. Confounds due to group differences were mitigated by matching a sub-sample of controls to glaucoma subjects. A convolutional neural network (CNN) accurately classified whether a subject has glaucoma using information from the primary visual connection to cortex (the optic radiations, OR), but not from non-visual brain connections. On the other hand, regularized linear regression could not classify glaucoma, and the CNN did not generalize to classification of age-group or of age-related macular degeneration. This suggests a unique non-linear signature of glaucoma in OR tissue properties.

neuroscience↗

bioRxiv2022

Optic radiations representing different eccentricities age differently

The white matter pathways that carry information from the foveal, macular, and peripheral visual fields have distinct biological properties. The optic radiations (OR) carry foveal and peripheral information from the thalamus to the primary visual cortex (V1) through adjacent but separate pathways in the white matter. Here, we perform white matter tractometry using pyAFQ 1 on a large sample of diffusion MRI (dMRI) data from the UK Biobank dataset (UKBB; N=6021; age 45-81). We use pyAFQ to characterize white matter tissue properties in parts of the OR that transmit information about the foveal, macular, and peripheral visual fields, and to characterize the changes in these tissue properties with age. We find that (1) independent of age there is higher fractional anisotropy, lower mean diffusivity, and higher mean kurtosis in the foveal and macular OR than in peripheral OR, consistent with denser, more organized nerve fiber populations in foveal/parafoveal pathways, and (2) age is associated with increased diffusivity and decreased anisotropy and kurtosis, consistent with decreased density and tissue organization with adulthood aging. However, anisotropy in foveal OR decreases faster with age than in peripheral OR, while diffusivity increases faster in peripheral OR, suggesting foveal/peri-foveal OR and peripheral OR differ in how they age.

neuroscience↗

bioRxiv2022

Development of the alpha rhythm is linked to visual white matter pathways and visual detection performance

Alpha is the strongest electrophysiological rhythm in awake humans at rest. Despite its predominance in the EEG signal, strong variations can be observed in alpha properties during development, with an increase of alpha frequency over childhood and adulthood. Here we tested the hypothesis that these changes of alpha rhythm are related to the maturation of visual white matter pathways. We capitalized on a large dMRI-EEG dataset (dMRI n=2,747, EEG n=2,561) of children and adolescents (age range: 5-21 years old) and showed that maturation of the optic radiation specifically accounts for developmental changes of alpha frequency. Behavioral analyses also confirmed that variations of alpha frequency are related to maturational changes in visual perception. The present findings demonstrate the close link between developmental variations in white matter tissue properties, electrophysiological responses, and behavior.

neuroscience↗

bioRxiv2022

Diffusion MRI Head Motion Correction Methods are Highly Accurate but Impacted by Denoising and Sampling Scheme

Correcting head motion artifacts in diffusion-weighted MRI (dMRI) scans is particularly challenging due to the dramatic changes in image contrast at different gradient strengths and directions. Head motion correction is typically performed using a Gaussian Process model implemented in FSLs Eddy. Recently, the 3dSHORE-based SHORELine method was introduced to correct any dMRI sequence that has more than one shell. Here we perform a comprehensive evaluation of both methods on realistic simulations of a software fiber phantom that provides known ground-truth head motion. We demonstrate that both methods perform remarkably well, but that performance can be impacted by sampling scheme, the pervasiveness of head motion, and the denoising strategy applied before head motion correction. Our study also provides an open and fully-reproducible workflow that could be used to accelerate evaluation studies of other dMRI processing methods in the future. HIGHLIGHTSO_LIBoth Eddy and SHORELine head motion correction methods performed quite well on a large variety of simulated data C_LIO_LIDenoising with MP-PCA can improve head motion correction performance when Eddy is used C_LIO_LISHORELine effectively corrects motion in non-shelled acquisitions C_LI

neuroscience↗

bioRxiv2022

White matter and literacy: a dynamic system in flux

Cross-sectional studies have linked differences in white matter tissue properties to reading skills. However, past studies have reported a range of, sometimes conflicting, results. Some studies suggest that white matter properties act as individual-level traits predictive of reading skill, whereas others suggest that reading skill and white matter develop as a function of an individuals educational experience. In the present study, we tested two hypotheses: a) that diffusion properties of the white matter reflect stable brain characteristics that relate to reading skills over development or b) that white matter is a dynamic system, linked with learning over time. To answer these questions, we examined the relationship between white matter and reading in a five-year longitudinal dataset and a series of large-scale, single-observation, cross-sectional datasets (N=14,249 total participants). We find that gains in reading skill correspond to longitudinal changes in the white matter. However, in the single-observation datasets, we find no evidence for the hypothesis that individual differences in white matter predict reading skill. These findings highlight the link between dynamic processes in the white matter and learning.

neuroscience↗