Biology subjects

Mahjoory, K.

Publications and source records attributed to Mahjoory, K..

5 recordsLinked to original sources

bioRxiv2026

Interpretable Hierarchical RNNs for rs-fMRI: Promise and Limits of Individualized Brain Dynamics

Modeling individual brain dynamics from resting-state fMRI (rs-fMRI) remains challenging due to substantial inter-subject variability, noise, and limited data length per subject. Here, we systematically evaluate whether hierarchical shallow piecewise-linear recurrent neural networks (shPLRNNs), recently introduced as interpretable dynamical system reconstruction models, can generate individualized rs-fMRI time series while preserving subject-specific functional connectivity structure. We applied the framework to 1,423 rs-fMRI samples from healthy participants of the Marburg-Munster Affective Disorders Cohort Study (MACS). Simulated rs-fMRI data reproduced substantial empirical FC structure, with comparable reconstruction accuracy on the validation and held-out test sets. Generalization to unseen individuals was heterogeneous and strongly depended on how typical a subjects connectivity pattern was relative to the training cohort, with template similarity explaining 37% of variance in reconstruction accuracy. Learned subject-specific parameters exhibited significant test-retest stability and higher within-subject than between-subject similarity on longitudinal data from two different timepoints, supporting their interpretation as individualized dynamical markers. Associations between individual parameters and demographic or cognitive variables were statistically significant but modest in effect size, and predictive performance remained below that obtained using empirical rs-fMRI features directly. Empirical FC was used as a reference for static subject information rather than as a target to be outperformed. Together, these results suggest that hierarchical shPLRNNs can extract meaningful and partially stable individual-specific dynamical structure from rs-fMRI data. The findings delineate key trade-offs between model expressivity, generalization and subject specificity, and point to directions for future methodological refinement in individualized brain modeling. Graphical AbstractA hierarchical dynamical RNN captures substantial individual rs-fMRI functional connectivity structure using compact subject-specific parameters embedded in shared population dynamics. The resulting representations generalize to held-out subjects and show test-retest stability, but only modest associations with phenotypic variables. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=150 SRC="FIGDIR/small/713153v3_ufig1.gif" ALT="Figure 1"> View larger version (61K): org.highwire.dtl.DTLVardef@1b029edorg.highwire.dtl.DTLVardef@90bdd6org.highwire.dtl.DTLVardef@9f6b77org.highwire.dtl.DTLVardef@486c9e_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2023

Convolutional neural networks can identify brain interactions involved in decoding spatial auditory attention

Human listeners have the ability to direct their attention to a single speaker in a multi-talker environment. The neural correlates of selective attention can be decoded from a single trial of electroencephalography (EEG) data. In this study, leveraging the source-reconstructed and anatomically-resolved EEG data as inputs, we sought to employ CNN as an interpretable model to uncover task-specific interactions between brain regions, rather than simply to utilize it as a black box decoder. To this end, our CNN model was specifically designed to learn pairwise interaction representations for 10 cortical regions from five-second input trials. By exclusively utilizing these features for decoding, our model was able to attain a median accuracy of 77.56% for within-participant and 65.14% for cross-participant classification. Through ablation analysis together with dissecting the features of the models and applying cluster analysis, we were able to discern the presence of alpha-band-dominated inter-hemisphere interactions, as well as alpha-and beta-band dominant interactions that were either hemisphere-specific or were characterized by a contrasting pattern between the right and left hemispheres. These interactions were more pronounced in parietal and central regions for within-participant decoding, but in parietal, central, and partly frontal regions for cross-participant decoding. These findings demonstrate that our CNN model can effectively utilize features known to be important in auditory attention tasks and suggest that the application of domain knowledge inspired CNNs on source-reconstructed EEG data can offer a novel computational framework for studying task-relevant brain interactions.

neuroscience↗

bioRxiv2019

The frequency gradient of human resting-state brain oscillations follows cortical hierarchies

The human cortex is characterized by local morphological features such as cortical thickness, myelin content and gene expression that change along the posterior-anterior axis. We investigated if these structural gradients are associated with a similar gradient in a prominent feature of brain activity - namely the frequency of brain oscillations. In resting-state MEG recordings from healthy participants (N=187), we found that the strongest peak frequency in a brain area decreases significantly, gradually and robustly along the posterior-anterior axis following the global hierarchy from early sensory to higher-order areas. This spatial gradient of peak frequency was significantly anticorrelated with the cortical thickness of corresponding areas representing a proxy of the cortical hierarchical level. This result indicates that the intrinsic resonance frequency decreases systematically from early sensory to higher-order areas and establishes a new structure-function relationship pertaining to brain oscillations as a core organizational principle that may underlie hierarchical specialization in the brain.

neuroscience↗

bioRxiv2019

Hemispheric asymmetries in resting-state EEG and fMRI are related to approach and avoidance behaviour, but not to eating behaviour or BMI

Much of our behaviour is driven by two motivational dimensions - approach and avoidance. These have been related to frontal hemispheric asymmetries in clinical and resting-state EEG studies: approach was linked to higher activity of the left relative to the right hemisphere, while avoidance was related to the opposite pattern. Increased approach behaviour, specifically towards unhealthy foods, is also observed in obesity and has been linked to asymmetry in the framework of the right-brain hypothesis of obesity. Here, we aimed to replicate previous EEG findings of hemispheric asymmetries for self-reported approach/avoidance behaviour and to relate them to eating behaviour. Further, we assessed whether resting fMRI hemispheric asymmetries can be detected and whether they are related to approach/avoidance, eating behaviour, and BMI. We analysed 3 samples: Sample 1 (n=117) containing EEG and fMRI data from lean participants, and Samples 2 (n=89) and 3 (n=152) containing fMRI data from lean, overweight, and obese participants. While in Sample 1 approach in women was related to EEG and fMRI hemispheric asymmetries, in Samples 2 and 3 this effect was not significant. Here, hemispheric asymmetries were neither related to BMI nor eating behaviour. Our study partly replicates previous EEG findings regarding hemispheric asymmetries and indicates that this relationship could also be captured using fMRI. Our findings suggest that eating behaviour and obesity are likely to be mediated by mechanisms not directly relating to frontal asymmetries in neuronal activation quantified with EEG and fMRI.

neuroscience↗

bioRxiv2019

BOLD and EEG Signal Variability at Rest Differently Relate to Aging in the Human Brain

Variability of neural activity is regarded as a crucial feature of healthy brain function, and several neuroimaging approaches have been employed to assess it noninvasively. Studies on the variability of both evoked brain response and spontaneous brain signals have shown remarkable changes with aging but it is unclear if the different measures of brain signal variability - identified with either hemodynamic or electrophysiological methods - reflect the same underlying physiology. In this study, we aimed to explore age differences of spontaneous brain signal variability with two different imaging modalities (EEG, fMRI) in healthy younger (25{+/-}3 years, N=135) and older (67{+/-}4 years, N=54) adults. Consistent with the previous studies, we found lower blood oxygenation level dependent (BOLD) variability in the older subjects as well as less signal variability in the amplitude of low-frequency oscillations (1-12 Hz), measured in source space. These age-related reductions were mostly observed in the areas that overlap with the default mode network. Moreover, age-related increases of variability in the amplitude of beta-band frequency EEG oscillations (15-25 Hz) were seen predominantly in temporal brain regions. There were significant sex differences in EEG signal variability in various brain regions while no significant sex differences were observed in BOLD signal variability. Bivariate and multivariate correlation analyses revealed no significant associations between EEG- and fMRI-based variability measures. In summary, we show that both BOLD and EEG signal variability reflect aging-related processes but are likely to be dominated by different physiological origins, which relate differentially to age and sex.

neuroscience↗