Frontiers in Neuroinformatics最新文献

We show that recent (mid-to-late 2024) commercial large language models (LLMs) are capable of good quality metadata extraction and annotation with very little work on the part of investigators for several exemplar real-world annotation tasks in the neuroimaging literature. We investigated the GPT-4o LLM from OpenAI which performed comparably with several groups of specially trained and supervised human annotators. The LLM achieves similar performance to humans, between 0.91 and 0.97 on zero-shot prompts without feedback to the LLM. Reviewing the disagreements between LLM and gold standard human annotations we note that actual LLM errors are comparable to human errors in most cases, and in many cases these disagreements are not errors. Based on the specific types of annotations we tested, with exceptionally reviewed gold-standard correct values, the LLM performance is usable for metadata annotation at scale. We encourage other research groups to develop and make available more specialized "micro-benchmarks," like the ones we provide here, for testing both LLMs, and more complex agent systems annotation performance in real-world metadata annotation tasks.

Electroencephalogram (EEG) signal analysis plays a vital role in diagnosing and monitoring alcoholism, where accurate classification of individuals into alcoholic and control groups is essential. However, the inherent noise and complexity of EEG signals pose significant challenges. This study investigates the impact of three signal denoising techniques' Discrete Wavelet Transform(DWT), Discrete Fourier Transform(DFT), and Discrete Cosine Transform (DCT) Non EEG signal classification performance. The motivation behind this study is to identify the most effective preprocessing method for enhancing deep learning model performance in this domain. A novel DWT-CNN-BiGRU model is proposed, which leverages CNN layers for spatial feature extraction and BiGRU layers for capturing temporal dependencies. Experimental results show that the DWT-based approach, combined with standard scaling, achieves the highest accuracy of 94%, with a precision of 0.94, a recall of 0.95, and an F1-score of 0.94. Compared to the baseline DWT-CNN-BiLSTM model, the proposed method provides a modest yet meaningful improvement of approximately 17% in classification accuracy. These findings highlight the superiority of DWT as a preprocessing method and validate the proposed model's effectiveness for EEG-based classification, contributing to the development of more reliable medical diagnostic tools.

Introduction: Understanding the cognitive phenotypes of elite athletes offers a unique perspective on the intricate interplay between neurological traits and high-performance behaviors. This study aligns with advancing neuroinformatics by proposing a novel framework designed to capture and analyze the multi-dimensional dependencies of cognitive phenotypes using systems neuroscience methodologies. Traditional approaches often face limitations in disentangling the latent factors influencing cognitive variability or in preserving interpretable data structures.

Methods: To address these challenges, we developed the Latent Cognitive Embedding Network (LCEN), an innovative model that combines biologically inspired constraints with state-of-the-art neural architectures. The model features a specialized embedding mechanism for disentangling latent factors and a tailored optimization strategy incorporating domain-specific priors and regularization techniques.

Results: Experimental evaluations demonstrate LCEN's superiority in predicting and interpreting cognitive phenotypes across diverse datasets, providing deeper insights into the neural underpinnings of elite performance.

Discussion: This work bridges computational modeling, neuroscience, and psychology, contributing to the broader understanding of cognitive variability in specialized populations.

Introduction: Mental health care is undermined by fragmented data collection, as incomplete datasets can compromise treatment efficacy and research. The BrainHealth Databank (BHDB) at the Centre for Addiction and Mental Health (CAMH) establishes the governance and infrastructure for a Learning Mental Health System that integrates digital tools, measurement-based care, artificial intelligence (AI), and open science to deliver personalized, data-driven care.

Methods: Central to the BHDB's approach is its comprehensive governance framework, which actively engages clinicians, researchers, data scientists, privacy and ethics experts, and patient and family partners. This codesigned approach ensures that digital health technologies are deployed ethically, securely, and effectively within clinical settings.

Results: By aligning data collection with clinical and research goals and harmonizing over 12 million data points from 33,000 patient trajectories, the BHDB enhances data quality, enables real-time decision support, and fosters continuous improvement.

Discussion: The BHDB provides a model for integrating AI and digital tools into mental health care, as well as research data collection, analyses, storage, and sharing through the BHDB Portal (https://bhdb.camh.ca).

Introduction: Enhancing the command capacity of motor imagery (MI)-based brain-computer interfaces (BCIs) remains a significant challenge in neuroinformatics, especially for real-world assistive applications. This study explores a multiclass BCI system designed to classify multiple MI tasks using a low-cost EEG device.

Methods: A BCI system was developed to classify six mental states: resting state, left and right hand movement imagery, tongue movement, and left and right lateral bending, using EEG data collected with the Emotiv EPOC X headset. Seven participants underwent a body awareness training protocol integrating mindfulness and physical exercises to improve MI performance. Machine learning techniques were applied to extract discriminative features from the EEG signals.

Results: Post-training assessments indicated modest improvements in participants' MI proficiency. However, classification performance was limited due to inter- and intra-subject signal variability and the technical constraints of the consumer-grade EEG hardware.

Discussion: These findings highlight the value of combining user training with MI-based BCIs and the need to optimize signal quality for reliable performance. The results support the feasibility of scalable, multiclass MI paradigms in low-cost, user-centered neurotechnology applications, while pointing to critical areas for future system enhancement.

Computational-neuroscience simulators have traditionally been constrained by tightly coupled simulation engines and modeling languages, limiting their flexibility and scalability. Retrofitting these platforms to accommodate new backends is often costly, and sharing models across simulators remains cumbersome. This paper puts forward an alternative approach based on the EDEN neural simulator, which introduces a modular stack that decouples abstract model descriptions from execution. This architecture enhances flexibility and extensibility by enabling seamless integration of multiple backends, including hardware accelerators, without extensive reprogramming. Through the use of NeuroML, simulation developers can focus on high-performance execution, while model users benefit from improved portability without the need to implement custom simulation engines. Additionally, the proposed method for incorporating arbitrary simulation platforms-from model-optimized code kernels to custom hardware devices-as backends offers a more sustainable and adaptable framework for the computational-neuroscience community. The effectiveness of EDEN's approach is demonstrated by integrating two distinct backends: flexHH, an FPGA-based accelerator for extended Hodgkin-Huxley networks, and SpiNNaker, the well-known, neuromorphic platform for large-scale spiking neural networks. Experimental results show that EDEN integrates the different backends with minimal effort while maintaining competitive performance, reaffirming it as a robust, extensible platform that advances the design paradigm for neural simulators by achieving high generality, performance, and usability.

Neuron reconstruction is a critical step in quantifying neuronal structures from imaging data. Advances in molecular labeling techniques and optical imaging technologies have spurred extensive research into the patterns of long-range neuronal projections. However, mapping these projections incurs significant costs, as large-scale reconstruction of individual axonal arbors remains time-consuming. In this study, we present a dataset comprising axon imaging volumes along with corresponding annotations to facilitate the evaluation and development of axon reconstruction algorithms. This dataset, derived from 11 mouse brain samples imaged using fluorescence micro-optical sectioning tomography, contains carefully selected 852 volume images sized at 192 × 192 × 192 voxels. These images exhibit substantial variations in terms of axon density, image intensity, and signal-to-noise ratios, even within localized regions. Conventional methods often struggle when processing such complex data. To address these challenges, we propose a distance field-supervised segmentation network designed to enhance image signals effectively. Our results demonstrate significantly improved axon detection rates across both state-of-the-art and traditional methodologies. The released dataset and benchmark algorithm provide a data foundation for advancing novel axon reconstruction methods and are valuable for accelerating the reconstruction of long-range axonal projections.

[This corrects the article DOI: 10.3389/fninf.2023.1244347.].

Introduction: Mathematical models serve as essential tools to investigate brain aging, the onset of Alzheimer's disease (AD) and its progression. By studying the representation of the complex dynamics of brain aging processes, such as amyloid beta (Aβ) deposition, tau tangles, neuro-inflammation, and neuronal death. Sensitivity analyses provide a powerful framework for identifying the underlying mechanisms that drive disease progression. In this study, we present the first local sensitivity analysis of a recent and comprehensive multiscale ODE-based model of Alzheimer's Disease (AD) that originates from our group. As such, it is one of the most complex model that captures the multifactorial nature of AD, incorporating neuronal, pathological, and inflammatory processes at the nano, micro and macro scales. This detailed framework enables realistic simulation of disease progression and identification of key biological parameters that influence system behavior. Our analysis identifies the key drivers of disease progression across patient profiles, providing insight into targeted therapeutic strategies.

Methods: We investigated a recent ODE-based model composed of 19 variables and 75 parameters, developed by our group, to study Alzheimer's disease dynamics. We performed single- and paired-parameter sensitivity analyses, focusing on three key outcomes: neural density, amyloid beta plaques, and tau proteins.

Results: Our findings suggest that the parameters related to glucose and insulin regulation could play an important role in neurodegeneration and cognitive decline. Second, the parameters that have the most important impact on cognitive decline are not completely the same depending on sex and APOE status.

Discussion: These results underscore the importance of incorporating a multifactorial approach tailored to demographic characteristics when considering strategies for AD treatment. This approach is essential to identify the factors that contribute significantly to neural loss and AD progression.

This study demonstrates the effectiveness of integrating cloud computing platforms with Course-based Undergraduate Research Experiences (CUREs) to broaden access to neuroscience education. Over four consecutive spring semesters (2021-2024), a total of 42 undergraduate students at Lawrence Technological University participated in computational neuroscience CUREs using brainlife.io, a cloud-computing platform. Students conducted anatomical and functional brain imaging analyses on openly available datasets, testing original hypotheses about brain structure variations. The program evolved from initial data processing to hypothesis-driven research exploring the influence of age, gender, and pathology on brain structures. By combining open science and big data within a user-friendly cloud environment, the CURE model provided hands-on, problem-based learning to students with limited prior knowledge. This approach addressed key limitations of traditional undergraduate research experiences, including scalability, early exposure, and inclusivity. Students consistently worked with MRI datasets, focusing on volumetric analysis of brain structures, and developed scientific communication skills by presenting findings at annual research days. The success of this program demonstrates its potential to democratize neuroscience education, enabling advanced research without extensive laboratory facilities or prior experience, and promoting original undergraduate research using real-world datasets.