Biology subjects

Sotero, R. C.

Publications and source records attributed to Sotero, R. C..

9 recordsLinked to original sources

bioRxiv2026

Beyond Mean Pooling: Statistical Meta-Aggregation for Subject-Level EEG Classification in Parkinson's Disease

Electroencephalography (EEG) provides a non-invasive, temporally precise tool for studying neural alterations associated with Parkinson's disease (PD). When labels are defined at the subject level but inputs are epoch level, a key design choice is how epoch-wise predictions are aggregated to a single subject-level decision. Most prior work uses mean pooling, which assumes all epochs contribute equally and independently. This study proposes a statistical meta-aggregation strategy that represents each subject by 14 descriptors of the epoch-wise prediction distribution, including measures of central tendency, dispersion, shape, and high-confidence concentration. Using a self-supervised contrastive encoder trained on task-evoked EEG from the Simon conflict task, and a sample of 90 subjects including 57 PD and 33 control, this work evaluates the proposed framework under subject-wise five-fold cross-validation. Replacing mean pooling with statistical meta-aggregation lifts subject-level ROC-AUC from 0.655 to 0.872 for logistic regression and from 0.724 to 0.840 for support vector machines, without modifying the encoder or the classifier family. Averaging predictions across three independent self-supervised pretraining seeds yields competitive performance and improved balanced accuracy. These results suggest that, in this dataset, the way epoch-level predictions are summarized at the subject level has a substantial effect on final PD classification performance.

neuroscience↗

bioRxiv2026

Physics-informed neural networks for subject-specific recovery of travelling-wave structure in resting-state fMRI

Resting-state BOLD fluctuations show reproducible spatiotemporal propagation that is not accounted for by conventional zero-lag functional connectivity. Travelling wave analyses based on phase gradients or statistical decomposition have problems when multiple wave generators overlap or when wave origins have to be localized. This work introduces a wave-informed physics-informed neural network that represents the parcellated BOLD signal as a superposition of seven spatially extended damped travelling wave generators, each periodically re-excited at its own characteristic frequency and fixed at initialization to one of the seven canonical Schaefer resting-state networks. A soft anatomical prior derived from each subject's diffusion-tensor structural connectivity is included to encode the hypothesis that white-matter pathways constrain wave direction. The pipeline is validated on synthetic data by three simulation studies, creating the operating envelope, and then applied to 41 cognitively normal participants of the ADNI 3 cohort. A within-subject ablation showed that mean alignment between recovered wave directions and underlying white matter geometry increased by a small but statistically detectable amount when using the structural prior (one-sidedWilcoxon signed-rank, p = 0.026), with exploratory network-level effects most evident in the visual, dorsal-attention, and salience/ventral-attention networks. The contribution is a network-anchored parametric inverse method returning subject-specific propagation descriptors from short resting-state scans.

neuroscience↗

bioRxiv2026

Identifying Network Factors Driving Changes in Brain Connectivity During Aging via Stochastic Actor-Oriented Models

Brain network organization emerges from ongoing network dynamics that continuously create, maintain, and reorganize connectivity patterns. At both the structural and functional levels, these dynamics shape network evolution throughout the lifespan. Here, we introduce a data-driven framework based on Stochastic Actor-Oriented Models (SAOMs), which model longitudinal network changes as a process of local edge formation and dissolution driven by local-connectivity factors. Applied to functional and structural brain networks from an aging cohort, the resulting models recovered a set of interacting mechanisms that collectively promote hub formation, short-range connectivity, and modular organization. These factors were derived through a novel data-driven factor-selection procedure rather than specified a priori, allowing network mechanisms underlying longitudinal change to be identified directly from observations. Functional networks exhibited substantially higher rates of change than structural networks, suggesting more frequent opportunities for network maintenance and reorganization. Structural networks additionally showed despecification in the hub-forming mechanisms with cognitive impairment. These results support a view of brain network organization as an actively maintained process emerging from ongoing local connectivity dynamics. More broadly, this framework provides a general approach for studying longitudinal network dynamics through the estimation of network-generating mechanisms over time.

neuroscience↗

bioRxiv2024

Parameter Estimation in Brain Dynamics Models from Resting-State fMRI Data using Physics-Informed Neural Networks

Conventional modeling of the Blood-Oxygen-Level-Dependent (BOLD) signal in resting-state functional Magnetic Resonance Imaging (rsfMRI) struggle with parameter estimation due to the complexity of brain dynamics. This study introduces a novel brain dynamics model (BDM) that directly captures BOLD signal variations through differential equations. Unlike dynamic causal models or neural mass models, we integrate hemodynamic responses into the signal dynamics, considering both direct and network-mediated neuronal activity effects. We utilize Physics-Informed Neural Networks (PINNs) to estimate the parameters of this BDM, leveraging their ability to embed physical laws into the learning process. This approach simplifies computational demands and increases robustness against data noise, providing a comprehensive tool for analyzing rsfMRI data. Leveraging the functional connectivity matrices scaled by the estimated parameters, we apply a state-of-the-art community detection method to elucidate the network structure. Our analysis reveals significant differences in the participation coefficients of specific brain regions when comparing neurotypical individuals to those with Autism Spectrum Disorder (ASD), with distinct patterns observed between male and female cohorts. These differences are consistent with regions implicated in previous studies, reinforcing the role of these areas in ASD. By integrating PINNs with advanced network analysis, we demonstrate a robust approach for dissecting the complex neural signatures of ASD, providing a promising direction for future research in neuroimaging and the broader field of computational neuroscience.

neuroscience↗

bioRxiv2023

Improving fMRI-based Autism Spectrum Disorder Classification with Random Walks-informed Feature Extraction and Selection

Functional magnetic resonance imaging (fMRI) is a non-invasive technique measuring brain activity by detecting blood flow changes, enabling the study of cognitive processes and brain states. However, the high dimensionality of resting-state (rs) fMRI data poses challenges for machine learning applications. Feature extraction (FE) and feature selection (FS) are critical for developing efficient machine learning models. Transforming raw data into meaningful features and selecting the most relevant ones, allows models to achieve improved generalization, accuracy, and robustness. Previous studies demonstrated the effectiveness of FE and FS methods for analyzing rs-fMRI data for Autism Spectrum Disorder (ASD) classification. In this study, we apply a random walks technique for correlation-based brain networks to extract features from rs-fMRI data, specifically the number of random walkers on each brain area. We then select significant features, i.e., brain areas with a statistically significant difference in the number of random walkers between neurotypical and ASD subjects. Our random walks-based FE and FS approach reduces the number of brain areas used in the classification and converts the functional connectivity matrix into a manageable vector, enabling faster computation. We examined 16 pipelines and tested support vector machines (SVM) and logistic regression for classification, identifying the optimal pipeline to consist of no filtering, no global signal regression (GSR), and FS, achieving a 76.54% classification accuracy with SVM. Our findings suggest that random walks capture a wide range of interactions and dynamics in brain networks, providing a deeper characterization of their structure and function, ultimately enhancing classification performance. CCS CONCEPTSComputing methodologies[->]Machine learning

neuroscience↗

bioRxiv2022

Deep EEG Source Localization via EMD-based fMRI High Spatial Frequency

Brain imaging with a high-spatiotemporal resolution is crucial for accurate brain-function mapping. Electroencephalography (EEG) and functional Magnetic Resonance Imaging (fMRI) are two popular neuroimaging modalities with complementary features that record brain function with high temporal and spatial resolution, respectively. One popular non-invasive way to obtain data with both high spatial and temporal resolutions is to combine the fMRI activation map and EEG data to improve the spatial resolution of the EEG source localization. However, using the whole fMRI map may cause spurious results for the EEG source localization, especially for deep brain regions. Considering the heads conductivity, deep regions sources with low activity are unlikely to be detected by the EEG electrodes at the scalp. In this study, we use fMRIs high spatial-frequency component to identify the local high-intensity activations that are most likely to be captured by the EEG. The 3D Empirical Mode Decomposition (3D-EMD), a data-driven method, is used to decompose the fMRI map into its spatial-frequency components. Different validation measurements for EEG source localization show improved performance for the EEG inverse-modeling informed by the fMRIs high-frequency spatial component compared to the fMRI-informed EEG source-localization methods. The level of improvement varies depending on the voxels intensity and their distribution. Our experimental results also support this conclusion.

neuroscience↗

bioRxiv2021

Holo-Hilbert Spectral-based Removal of Gamma-Band Noise from Simultaneous EEG-fMRI Recordings

Simultaneous EEG-fMRI is a growing and promising field, as it has great potential to further our understanding of the spatiotemporal dynamics of brain function in health and disease. In particular, there is much interest in understanding the fMRI correlates of brain activity in the gamma band (> 30 Hz), as these frequencies are thought to be associated with cognitive processes involving perception, attention, and memory, as well as with disorders such as schizophrenia and autism. However, progress in this area has been limited due to issues such as MR-induced artifacts in EEG recordings, which seem to be more problematic for gamma frequencies. This paper presents a noise removal method for the gamma band of EEG that is based on the Holo-Hilbert spectral analysis (HHSA), but with a new implementation strategy. HHSA uses a nested empirical mode decomposition (EMD) to identify amplitude and frequency modulations (AM and FM, respectively) by averaging over frequencies with high and significant powers. Our method examines gamma band by applying two layers of EMD to the FM and AM components, removing components with very low power based on the power-instantaneous frequency spectrum, and subsequently reconstructs the denoised gamma-band signal from the remaining components. Simulations demonstrate that our proposed method efficiently reduces artifacts while preserving the original gamma signal which is especially critical for simultaneous EEG/fMRI studies.

neuroscience↗

bioRxiv2019

A method for multiscale community detection in brain networks

The identification of community structure in graphs continues to attract great interest in several fields. Network neuroscience is particularly concerned with this problem considering the key roles communities play in brain processes and functionality. Most methods used for community detection in brain graphs are based on the maximization of a parameter-dependent modularity function that often obscures the physical meaning and hierarchical organization of the partitions of network nodes. In this work, we present a new method able to detect communities at different scales in a natural, unrestricted way. First, to obtain an estimation of the information flow in the network we release random walkers to freely move over it. The activity of the walkers is separated into oscillatory modes by using empirical mode decomposition. After grouping nodes by their co-occurrence at each time scale, k-modes clustering returns the desired partitions. Our algorithm was first tested on benchmark graphs with favorable performance. Next, it was applied to real and simulated anatomical and/or functional connectomes in the macaque and human brains. We found a clear hierarchical repertoire of community structures in both the anatomical and the functional networks. The observed partitions range from the evident division in two hemispheres -in which all processes are managed globally- to specialized communities seemingly shaped by physical proximity and shared function. Our results stimulate the research of hierarchical community organization in terms of temporal scales of information flow in the brain network. Highlights- Oscillatory modes of networks signals carry information on architectural rules. - Meaningful partitions of the brain networks are found over different temporal scales. - The multiscale organization of the brain responds to the function of its components.

neuroscience↗

bioRxiv2019

Estimation of Global and Local Complexities of Brain Networks: A Random Walks Approach

The complexity of brain activity has been observed at many spatial scales and there exists increasing evidence supporting its use in differentiating between mental states and disorders. Here we proposed a new measure of network (global) complexity that is constructed as the sum of the complexities of its nodes (i.e, local complexity). The local complexity of each node is regarded as an index that compares the sample entropy of the time series generated by the movement of a random walker on the network resulting from removing the node and its connections, with the sample entropy of the time series obtained from a regular lattice (the ordered state) and an Erdos-Renyi network (disordered state). We studied the complexity of fMRI-based resting-state functional networks. We found that positively correlated, or \"pos\", network comprising only the positive functional connections has higher complexity than the anticorrelation (\"neg\") network (comprising the negative functional connections) and the network consisting of the absolute value of all connections (\"abs\"). We also found a significant correlation between complexity and the strength of functional connectivity. For the pos network this correlation is significantly weaker at the local scale compared to the global scale, whereas for the neg network the link is stronger at the local scale than at the global scale, but still weaker than for the pos network. Our results suggest that the pos network is related to the information processing in the brain and should be used for functional connectivity analysis instead of the abs network as is usually done.

neuroscience↗