Frontiers in Computational Neuroscience最新文献

Accurate detection and segmentation of multiple sclerosis (MS) lesions in brain Magnetic Resonance Imaging (MRI) is a challenging task due to their small size, irregular shape, and variability in different imaging modalities. Precise segmentation of MS lesions from brain MRI is vital for early diagnosis, disease progression monitoring, and treatment planning. We introduce MS-DASPNet, a Dual Attention Guided Deep Neural Network specifically designed to address the challenges of MS lesion detection, including small lesion sizes, low contrast, and heterogeneous appearance. MS-DASPNet employs a VGG-16-based encoder, an Atrous Spatial Pyramid Pooling (ASPP) bottleneck for multi-scale context learning, and dual attention modules in each skip connection to simultaneously refine spatial details and enhance channel-wise feature representation. Evaluations on four publicly available datasets, namely ISBI-2015, Mendeley, MICCAI-2016, and MICCAI-2021, demonstrate that MS-DASPNet achieves superior Precision, Dice, Sensitivity, and Jaccard scores compared to state-of-the-art methods. MS-DASPNet attains a Dice score of 0.8736 on the MICCAI-2016 dataset and 0.8706 on the MICCAI-2021 dataset, both outperforming existing segmentation techniques, highlighting its robustness and effectiveness in accurate MS lesion segmentation.

Purpose: An information theory-based framework is proposed in attempt to explain insistence on sameness in autism as an instance of a general behavior pattern in which an individual tries to reduce surprise and uncertainty. It offers a new definition of autism as an impairment in which cognitive functions are restricted to discrimination, memorization and prediction of tangible properties of the environment.

Methods: An analogy between insistence on sameness and constrained minimization of the entropy metric is observed and examined for a set of assumptions that describe cognitive limitations of a person with autism. The metric is given by the formula D _H (R, M) = H(R|M)+H(M|R), where R represents sequences of random stimuli, M is a memory that stores and retrieves them, and where H(·|·) denotes their conditional entropies interpreted as surprise and uncertainty, respectively.

Results: It is first inferred that to minimize the metric an individual can learn about R (and store that knowledge in M) or can restrict R to the already known M. Then, it is concluded that insistence on sameness is a manifestation of the latter. Moreover, it is shown that the proposed framework: (1) Helps to quantify the concepts of surprise, uncertainty, sensory overload and deprivation, anxiety, comfort zone, disappointment, disorientation, pedantry, rigidness, observance or aberrant precision. (2) Leads to a list of guidelines for learning therapies and daily care routines, and allows them to be defined as optimization algorithms and implemented as programs for robotic live-in caregivers. (3) Can be validated with the help of a Turing test-like approach that requires no experiments involving individuals with autism.

Conclusion: The framework-if positively validated-will provide advantages of both theoretical and practical importance: it explains the insistent on sameness as a consequence of cognitive restrictions and offers formal foundations and design guidelines for therapies aimed at improving self-reliance of individuals with autism in basic activities of daily living.

The recently developed phenomenological renormalization group (pRG) analysis has uncovered scale-free properties in large-scale neural population recordings across recording modalities, including extracellular electrophysiology and calcium imaging. The convergence of these properties across the datasets hints at universal neural behavior. Yet, it is unknown how differences in temporal resolution and measurement details affect pRG scaling. Here, we use a network model known to produce scaling under pRG analysis as a testbed to assess how recording and analysis choices shape inferred scaling exponents. We show that scaling properties depend on the choices of temporal binning, measurement nonlinearities, and deconvolution, and that the quality of scaling for cluster covariance eigenvalues is particularly sensitive to measurement effects. Moreover, all scaling exponents shift substantially with these transformations, even when the underlying neural dynamics are identical. Together, these results show how experimental choices can change pRG scaling and provide a framework for separating scaling driven by neural dynamics from that introduced by the recording method.

Serotonin is thought to regulate emotional learning and memory, but there remains much to be explored regarding its causal role in cued fear conditioning and extinction (CFC-E). Recent in vivo recording of dorsal raphe nucleus serotonin neuronal activity during CFC-E paradigm showed that the time course of serotonin level includes both rapid responses to conditioned and unconditioned stimuli and a slowly accumulating component that spans inter-trial intervals and reverses during extinction. By reviewing the studies that directly link the fear expression during CFC-E to the acute or chronic perturbations of serotonin dynamics at the organism level or within specific brain areas via pharmacological, genetic, and projection-specific manipulations, we argue that theoretical models defining the causal role of serotonin must incorporate continuous-time serotonin dynamics.

Introduction: Probabilistic Stimulation Maps (PSMs) are increasingly employed to identify brain regions associated with optimal therapeutic outcomes in Deep Brain Stimulation (DBS). However, their reliability and generalizability are challenged by the limited size of most patient cohorts and the inherent variability introduced by different statistical methods and input data configurations. This study aimed to investigate the geometrical variability of Probabilistic Sweet Spots (PSS) as a function of both the number of patients (nPat) and the number of stimulations per patient (nStim), and to model a stability boundary defining the minimum data requirements for obtaining geometrically stable PSS.

Methods: Three statistical approaches-Bayesian t-test, Wilcoxon test with False Discovery Rate (FDR) correction, and Wilcoxon test with nonparametric permutation correction-were applied to two patient cohorts: a primary cohort of 36 patients undergoing DBS for Parkinson's Disease (PD), and a secondary cohort of 61 patients treated for Essential Tremor (ET), used to assess generalizability. Stimulation test data was collected intra-operatively for the first cohort and post-operatively for the second one. Geometric stability was evaluated based on variability in PSS volume extent and centroid location.

Results: The analysis revealed a non-linear trade-off between nPat and nStim to yield stable PSS. A stability boundary was defined, representing the minimum combinations of nPat-nStim required for anatomically robust PSS. Among the tested methods, the Bayesian t-test achieved stability with smaller sample sizes (∼15 patients) and demonstrated a consistent performance across both cohorts. In contrast, the Wilcoxon-based methods showed variable behavior between cohorts, which differed in symptom type and testing phase (intra-operative testing vs. post-operative screening).

Discussion: The proposed PSS stability boundary provides a practical reference for designing DBS studies and stimulation screening protocols aimed at probabilistic mapping. The Bayesian t-test emerged as a reliable method across both cohorts, supporting its potential in studies with limited sample sizes and scenarios where the method needs to be readily generalized to varying symptoms. These findings underscore the importance of considering both cohort size and stimulation count in probabilistic DBS mapping and call for further investigation into method-specific sensitivities to clinical and procedural factors.

Community detection is a crucial task in network research, applicable to social systems, biology, cybersecurity, and knowledge graphs. Recent advancements in graph neural networks (GNNs) have exhibited significant representational capability; yet, they frequently experience instability and erroneous clustering, often referred to as "hallucinations." These artifacts stem from sensitivity to high-frequency eigenmodes, over-parameterization, and noise amplification, undermining the robustness of learned communities. To mitigate these constraints, we present F²-CommNet, a Fourier-Fractional neural framework that incorporates fractional-order dynamics, spectrum filtering, and Lyapunov-based stability analysis. The fractional operator implements long-memory dampening that mitigates oscillations, whereas Fourier spectral projections selectively attenuate eigenmodes susceptible to hallucination. Theoretical analysis delineates certain stability criteria under Lipschitz non-linearities and constrained disturbances, resulting in a demonstrable expansion of the Lyapunov margin. Experimental validation on synthetic and actual networks indicates that F²-CommNet reliably diminishes hallucination indices, enhances stability margins, and produces interpretable communities in comparison to integer-order GNN baselines. This study integrates fractional calculus, spectral graph theory, and neural network dynamics, providing a systematic method for hallucination-resistant community discovery.

Extracting features from abnormal brain regions in schizophrenia patients' brain images holds significant importance for aiding diagnosis. However, existing methods remained limited in simultaneously capturing spatiotemporal information. Dynamic mode decomposition (DMD) effectively extracts spatiotemporal features from dynamic systems, making it suitable for time-series signals such as functional magnetic resonance imaging (fMRI) and electrocorticography (ECoG). This study utilized resting-state fMRI data from 68 healthy subjects and 68 schizophrenia patients. The DMD method was employed to extract the mean amplitude of dynamic patterns as features, with feature selection conducted via Least Absolute Shrinkage and Selection Operator (LASSO) regression. A support vector machine (SVM) was further employed to validate the predictive capability of the selected features across subject groups. Based on the LASSO screening, we identified brain regions exhibiting significant inter-group differences in mean amplitude, designated these as abnormal regions, and subsequently analyzed their functional deviations. The DMD method not only provided explicit temporal dynamic representations of brain activity but also supported signal reconstruction and prediction, thereby enhancing feature interpretability. Results demonstrated that DMD effectively extracted mean amplitude features from fMRI data. Combined with LASSO and SVM, it enabled the identification of abnormal brain regions and functional abnormalities in schizophrenia patients. Furthermore, this method captured frequency-dependent signal patterns, with extracted features correlating with both regional activation intensity and functional connectivity. This approach provides novel insights for exploring potential biomarkers of psychiatric disorders.

Childhood absence epilepsy (CAE) often resolves during adolescence, a period marked by hormonal and neurosteroid changes associated with puberty. However, remission does not occur in all individuals. To investigate this clinical heterogeneity, we developed a simplified thalamocortical model with a layered cortical structure, using deep-layer intrinsically bursting (IB) neurons to represent frontal cortex and regular spiking (RS) neurons modeling the parietal cortex. By simulating two cortical configurations, we explored how variations in neuronal composition and frontocortical connectivity influence seizure dynamics and the effectiveness of allopregnanolone (ALLO) in resolving pathological spike-wave discharges (SWDs) associated with CAE. While both models exhibited similar physiological and pathological oscillations, only the parietal-dominant network (with a higher proportion of RS neurons in layer 5) recovered from SWDs under increased frontocortical connectivity following ALLO administration. These findings suggest that neuronal composition critically modulates ALLO-mediated resolution of SWDs, providing a mechanistic link between structural connectivity and clinical outcomes in CAE, and highlighting the potential for personalized treatment strategies based on underlying network architecture.

Action potentials (AP) are the basic elements of information processing in the nervous system. Understanding AP generation mechanisms is a critical step to understand how neurons encode information. However, an individual neuron might fire APs with various shapes even in response to the same stimulus, and the mechanisms responsible for this variability remain unclear. Therefore, we analyzed four AP attributes including AP rapidity and threshold during consecutive bursts from three neuron types using intracellular electrophysiological recordings. In response to consecutive current steps, the AP attributes in evoked spike trains show two distinctive patterns across different neurons: (1) The first APs from each train always have comparable properties regardless of the stimulus strength; (2) The attributes of the subsequent APs during each pulse monotonically change during the burst, where the magnitude of AP attribute change during each pulse increases with increasing stimulation strength. Various conductance-based models were explored to determine if they replicated the observed AP bursts. The observed patterns could not be replicated using the classical HH-type models, or modified HH model with cooperative Na⁺ gating. However, adding ion concentration dynamics to the model reproduced the AP attribute variation, and the magnitude of change during a pulse correlated with change in dynamic reversal potential (DRP), but failed to replicate the first AP attributes pattern. Then, the role of cooperative Na⁺ gating on neuronal firing dynamics was investigated. Inclusion of cooperative gating restored the first APs' attributes and enhanced the magnitude of modeled variation of some AP attributes to better agree with observed data. We conclude that changes in local ion concentrations could be responsible for the monotonic change in APs attributes during neuronal bursts, and cooperative gating of Na⁺ channels can enhance the effect. Thus, the two mechanisms could contribute to the observed variability in neuronal response.