Neuroinformatics最新文献

Resting state functional magnetic resonance imaging (fMRI) is a useful technique to characterize functional connectivity patterns between regions of the brain, based on the Fisher-transformed Pearson correlations in the BOLD signal. Pinpointing how connectivity patterns change in neuropathies like early-stage psychosis (ESP) can help understand the disorders and track progression. Using study data from 21 ESP subjects with complete data for three consecutive scans, we examined connectivity changes throughout the whole brain with a region of interest (ROI) to ROI-based approach for ROI defined by the Harvard-Oxford cortical and subcortical atlases, supplemented by the AAL atlas for the cerebellum, and by networks defined by the CONN toolbox independent component analysis of the Human Connectome Project. We applied latent growth modelling, which is a type of structural equation modelling, to these connectivity measurements across baseline and follow-up visits. The models use age, community functioning, and negative symptoms at baselines as the covariates for subject-specific slope and intercept of the longitudinal measurements. After stringent thresholding cutoffs of root mean square error of approximation, standardized root mean square residual, comparative fit index, and Benjamini-Hochberg corrected p-value, we found a subset of connectivity measurements with significant longitudinal slopes (N = 18 atlas, N = 6 network), and used the subject's slope estimates to stratify these subjects into three clusters based on how the ROI-to-ROI correlations of functional connectivity change over time. The connections with significant slopes include atlas level regions like the temporal lobe, fronto-parietal lobe, and cerebellum, and network level patterns like the DMN, FPN, and Salience Networks. The structural equation modelling approach identifies ROIs whose functional connectivity changes over time, indicating the ROIs most dynamic during ESP. This highlights the utility of latent growth models for the analysis of longitudinal functional connectivity measures across the whole brain with relatively small sample sizes.

The amygdala plays a central role in emotion, memory, and decision-making and comprises approximately 13 distinct nuclei with connectivity. Despite its functional importance, high-resolution subnuclear mapping is challenging. This study aimed to construct a 3D model of the anatomical location of the amygdala in the brain and a functional dynamic model of the amygdala, integrating deep learning and elastic shape metrics. We used multimodal datasets from the Julich-Brain Atlas, BigBrain Project, and FreeSurfer, which were aligned with the Montreal Neurological Institute (MNI) and Colin 27 spaces. Subnuclei segmentation was performed using a Bayesian Fully Convolutional Network (FCN), and geometric morphometrics were analyzed using elastic shape analysis on the unit sphere. Functional dynamics were simulated using a MATLAB-based model of the amygdala incorporating theta (4-8 Hz) and gamma (30-40 Hz) oscillations with spike-timing-dependent plasticity (STDP). The mean MNI coordinates of the left and right amygdalae were (-20, -4, -15) and (22, -2, -15), respectively, with an inter-amygdalar distance of 42.48 mm. The Dice Similarity Coefficients (DSCs) for FCN-based subnuclear segmentation were as follows: basolateral amygdala (BLA) nucleus = 0.89 ± 0.03, centromedial nucleus = 0.83 ± 0.04, and cortical nucleus = 0.81 ± 0.05. Principal component analysis of elastic shape metrics revealed post-traumatic stress disorder (PTSD)-related morphological deviations, with the first principal component (PC1) accounting for 38% of the variance (p < 0.01). Oscillatory simulations captured the BLA rhythm dynamics and STDP-induced synaptic changes. This study presents a comprehensive 3D model of the human amygdala that bridges anatomical accuracy with computational modeling. Unlike prior models that focus solely on structural or functional domains, our approach integrates subnuclear segmentation, morphometrics, and real-time functional simulation. This study introduces a fully integrated anatomical-functional 3D model of the human amygdala, providing a translational platform for neuromodulation targeting, psychiatric diagnostics, and computational neuroengineering applications.

Multisensory integration is a fundamental neural mechanism crucial for understanding cognition. Multiple theoretical models exist to account for the computational processes underpinning this mechanism. However, there is an absence of a consolidated framework that facilitates the examination of multisensory integration across diverse experimental and computational contexts. We introduce Scikit-NeuroMSI, an accessible Python-based open-source framework designed to streamline the implementation and evaluation of computational models of multisensory integration. The capabilities of Scikit-NeuroMSI were demonstrated in enabling the implementation of multiple models of multisensory integration at different levels of analysis. Furthermore, we illustrate the utility of the software in systematically exploring the model's behavior in spatiotemporal causal inference tasks through parameter sweeps in simulations. Particularly, we conducted a comparative analysis of Bayesian and network models of multisensory integration to identify commonalities that may enable to bridge both levels of description, addressing a key research question within the field. We discuss the significance of this approach in generating computationally informed hypotheses in multisensory research. Recommendations for the improvement of this software and directions for future research using this framework are presented.

This study used voxel- and surface-based morphometry to analyze the changes in gray matter structure in MS patients and their correlation with clinical scales. An analysis was conducted on the structural magnetic resonance imaging data of 30 patients with MS who met the inclusion criteria and 30 healthy controls (HCs). Clinical disability was evaluated using the Expanded Disability Status Scale (EDSS) and the timed 25-foot walk test (T25FW). Cognitive function was assessed with the Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA). Psychiatric symptoms were measured via the Hamilton Anxiety Scale (HAMA) and Hamilton Depression Scale (HAMD). Imaging data were also collected from the MS and healthy control groups, and SPM12 and CAT12 analyzed the images. After controlling for age and gender, voxel- and surface-based morphometry were used to study inter-group differences. Finally, the discrepancy data were correlated with the clinical scales. Compared to the HC group, the gray matter volume reduction in the MS group was mainly concentrated in the deep gray matter, with a small portion located in the cortical gray matter (FWE-corrected p-value < 0.05). Cortical thickness was significantly reduced in multiple dispersed regions of the brain bilaterally in the MS group compared to hc (FWE-corrected p-value < 0.05), and there was no obvious anatomical connection between these regions. Correlation analyses revealed: A negative correlation between caudate nucleus volume and EDSS scores (R = -0.415, p = 0.031); a positive correlation between the right parahippocampal gyrus and HAMA scores (R = 0.392, p = 0.039); a positive correlations of the right postcentral gyrus with both MMSE (R = 0.433, p = 0.021) and MoCA scores (R = 0.431, p = 0.022); a positive correlation between the left paracentral lobule and MoCA scores (R = 0.389, p = 0.041). A pattern of multiple gray matter structural changes was identified in our study, and a clinical correlation between structural changes was found. Grey matter volume and cortical thickness hold substantial promise as markers of disease progression and have the potential to respond to neuroprotective treatments for MS neurodegeneration.

Non-invasive intracranial pressure (ICP) monitoring can help clinicians safely and efficiently monitor spaceflight-associated neuro-ocular syndrome (SANS), idiopathic intracranial hypertension, and traumatic brain injury in astronauts. Current invasive ICP measurement techniques are unsuitable for austere environments like spaceflight. In this study, we explore the potential of plasma-derived cell-free RNA (cfRNA) biomarkers as non-invasive alternatives to cerebrospinal fluid (CSF) markers for ICP assessment. We conducted a secondary analysis of NASA's Open Science Data Repository datasets 363-364, focusing on plasma and CSF biomarkers related to ICP and neurovascular health. An ensemble model combining Support Vector Machine, Gradient Boosting Regressor, and Ridge Regression was developed to capture plasma-CSF biomarker relationships. To address limited sample size, we employed a Generative Adversarial Network (GAN) to generate synthetic plasma-CSF biomarker pairs, expanding the dataset from 29 to 279 samples. The model's performance was evaluated using Mean Squared Error (MSE) and validated against real biomarker data. The GAN-augmented ensemble model achieved high predictive accuracy with an MSE of 0.0044. Synthetic plasma-CSF pairs closely aligned with actual biomarker distributions, demonstrating their effectiveness in reducing overfitting and enhancing model robustness. Strong correlations between plasma-derived RNA biomarkers and corresponding CSF indicators support their potential as non-invasive proxies for ICP assessment. This study establishes a novel framework for non-invasive ICP monitoring using plasma cfRNA profiles enriched with GAN-generated synthetic data. The approach shows promise for both spaceflight and clinical applications, potentially broadening diagnostic capabilities for ICP-related conditions. However, further validation across diverse populations is necessary, along with careful consideration of bioethical and data security issues associated with synthetic data use in clinical diagnostics.

Reproducibility of neuroimaging analyses and aggregation of heterogenous datasets are significant challenges in human subjects imaging research. This stems in part from a lack of an easy to use and universal data format that encompasses all steps of neuroimaging. The BIDS format has become widely adopted, however it is increasingly complex to implement as features are added, with the documentation now exceeding 500 pages. As such, there is a need for standards that can handle the complexity of the data while minimizing the complexity of the format. Here we present a simple but generalizable data sharing specification, called the squirrel format (not related to the squirrel programming language), to share imaging data in a simple, but flexible, specification. It is so named because squirrels are effective at storing significant quantities of food and knowing exactly where and when to find it. The design objectives of the format specification are to 1) store subject information, experimental parameters, raw data, analyzed data, and analysis methods 2) organize data in a human-readable hierarchy 3) enable easy sharing and dissemination of data packages. We developed a relational hierarchy with a structured representation of all steps of neuroimaging data collection and analysis, and a generalizable specification to store any modality of neuroimaging data, which satisfies the design objectives. Additionally, redundancy is minimized by using relational database principles. The specification allows all research data to be classified into one of ten object types, thus simplifying the sharing of neuroimaging data. Like how squirrels employ 'chunking', the squirrel format chunks data into a manageable number of object types. The squirrel format was developed to share neuroimaging data but can be generalized to share any imaging research.

Despite extensive cerebellar research, the functional role of individual cerebellar micro complexes (CmCs) in motor coordination remains debated. This study aimed to utilise a reductionist approach to model the CmC function in motor control using the Histologic Boolean Mapping (HBM-VNR) framework and validate it through replication of features observed in the literature. HBM-VNR modelled each neuron within the CmC as a Boolean expression derived from its architectural connectivity. The model incorporates the Variable Neuronal Response (VNR) synaptic model, introducing probabilistic post-synaptic firing to reflect physiological variability. Motor control dynamics follow the cerebellar brain inhibition phenomenon, where Deep Cerebellar Nucleus (DCN) firing activates the antagonist muscles. The model performed the task of feedback-control in an idealised joint following a desired sinusoidal position. HBM-VNR produced a minimalistic model that reproduced adaptive compensation to external forces and predicted intention tremor when CmC population was reduced, and the expected ethanol induced motor impairments. Simulated firing patterns of the DCN and Purkinje cell showed patterns resembling real recordings both in physiological and pathological situations. The Shifting Central Frequency Hypothesis (SCFH) was suggested to explain the CmC comparator functionality. This study presents HBM-VNR as a histologically grounded modelling approach for neural circuits. HBM-VNR simulated adaptive motor control and predicted neocerebellar syndrome symptomatology and alcohol intoxication effects. SCFH offers a computational mechanism consistent with the cerebellar internal model theories and places CmC as the basis for motor learning in line with the literature, positioning HBM-VNR as a scalable framework for neuroanatomical modelling.

Major Depressive Disorder (MDD) significantly impacts global health, impairing individual functioning and increasing socioeconomic burden. Developing innovative, interpretable approaches for its identification is essential for improving diagnosis and guiding treatment. This study introduces a novel framework designed to classify MDD using resting-state functional MRI (rs-fMRI) data. Our framework follows three stages: First, Node2Vec extracts rich, low-dimensional brain region embeddings from functional connectivity (FC) networks, capturing their complex topological information. Second, these informative embeddings feed a Graph Attention Network (GAT) which, via multi-head attention, identifies and weighs discriminative inter-regional functional connections, refining them into a potent graph representation. Third, these GAT-derived representations are classified by an ensemble classifier (Random Forest, SVM, MLP) for robust MDD identification. The model achieved classification accuracies of 78.73% and 92.94% on the REST-meta-MDD and SRPBS-MDD datasets, respectively. Moreover, the attention mechanism revealed that resting-state functional connectivity of regions within the Default Mode Network (DMN) and Frontoparietal Network (FPN) were among the most discriminative features for distinguishing MDD from healthy controls. The attention mechanism enhanced interpretability by highlighting significant brain regions linked to MDD. Compared to traditional methods, this GNN-based approach effectively captures complex brain connectivity patterns and offers improved interpretability, ultimately aiding healthcare professionals in diagnosing MDD more accurately.