Pub Date : 2024-09-09DOI: 10.3389/fncom.2024.1460006
Stephan Lochner, Daniel Honerkamp, Abhinav Valada, Andrew D. Straw
Bees are among the master navigators of the insect world. Despite impressive advances in robot navigation research, the performance of these insects is still unrivaled by any artificial system in terms of training efficiency and generalization capabilities, particularly considering the limited computational capacity. On the other hand, computational principles underlying these extraordinary feats are still only partially understood. The theoretical framework of reinforcement learning (RL) provides an ideal focal point to bring the two fields together for mutual benefit. In particular, we analyze and compare representations of space in robot and insect navigation models through the lens of RL, as the efficiency of insect navigation is likely rooted in an efficient and robust internal representation, linking retinotopic (egocentric) visual input with the geometry of the environment. While RL has long been at the core of robot navigation research, current computational theories of insect navigation are not commonly formulated within this framework, but largely as an associative learning process implemented in the insect brain, especially in the mushroom body (MB). Here we propose specific hypothetical components of the MB circuit that would enable the implementation of a certain class of relatively simple RL algorithms, capable of integrating distinct components of a navigation task, reminiscent of hierarchical RL models used in robot navigation. We discuss how current models of insect and robot navigation are exploring representations beyond classical, complete map-like representations, with spatial information being embedded in the respective latent representations to varying degrees.
{"title":"Reinforcement learning as a robotics-inspired framework for insect navigation: from spatial representations to neural implementation","authors":"Stephan Lochner, Daniel Honerkamp, Abhinav Valada, Andrew D. Straw","doi":"10.3389/fncom.2024.1460006","DOIUrl":"https://doi.org/10.3389/fncom.2024.1460006","url":null,"abstract":"Bees are among the master navigators of the insect world. Despite impressive advances in robot navigation research, the performance of these insects is still unrivaled by any artificial system in terms of training efficiency and generalization capabilities, particularly considering the limited computational capacity. On the other hand, computational principles underlying these extraordinary feats are still only partially understood. The theoretical framework of reinforcement learning (RL) provides an ideal focal point to bring the two fields together for mutual benefit. In particular, we analyze and compare representations of space in robot and insect navigation models through the lens of RL, as the efficiency of insect navigation is likely rooted in an efficient and robust internal representation, linking retinotopic (egocentric) visual input with the geometry of the environment. While RL has long been at the core of robot navigation research, current computational theories of insect navigation are not commonly formulated within this framework, but largely as an associative learning process implemented in the insect brain, especially in the mushroom body (MB). Here we propose specific hypothetical components of the MB circuit that would enable the implementation of a certain class of relatively simple RL algorithms, capable of integrating distinct components of a navigation task, reminiscent of hierarchical RL models used in robot navigation. We discuss how current models of insect and robot navigation are exploring representations beyond classical, complete map-like representations, with spatial information being embedded in the respective latent representations to varying degrees.","PeriodicalId":12363,"journal":{"name":"Frontiers in Computational Neuroscience","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142205486","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-09-04DOI: 10.3389/fncom.2024.1452001
Andreas Bahmer,Johanna M Rimmele,Daya Shankar Gupta
{"title":"Editorial: Understanding the role of oscillations, mutual information and synchronization in perception and action.","authors":"Andreas Bahmer,Johanna M Rimmele,Daya Shankar Gupta","doi":"10.3389/fncom.2024.1452001","DOIUrl":"https://doi.org/10.3389/fncom.2024.1452001","url":null,"abstract":"","PeriodicalId":12363,"journal":{"name":"Frontiers in Computational Neuroscience","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142260785","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-09-03DOI: 10.3389/fncom.2024.1388504
Kenneth Smith, Sharlee Climer
Late-onset Alzheimer disease (AD) is a highly complex disease with multiple subtypes, as demonstrated by its disparate risk factors, pathological manifestations, and clinical traits. Discovery of biomarkers to diagnose specific AD subtypes is a key step towards understanding biological mechanisms underlying this enigmatic disease, generating candidate drug targets, and selecting participants for drug trials. Popular statistical methods for evaluating candidate biomarkers, fold change (FC) and area under the receiver operating characteristic curve (AUC), were designed for homogeneous data and we demonstrate the inherent weaknesses of these approaches when used to evaluate subtypes representing less than half of the diseased cases. We introduce a unique evaluation metric that is based on the distribution of the values, rather than the magnitude of the values, to identify analytes that are associated with a subset of the diseased cases, thereby revealing potential biomarkers for subtypes. Our approach, Bimodality Coefficient Difference (BCD), computes the difference between the degrees of bimodality for the cases and controls. We demonstrate the effectiveness of our approach with large-scale synthetic data trials containing nearly perfect subtypes. In order to reveal novel AD biomarkers for heterogeneous subtypes, we applied BCD to gene expression data for 8,650 genes for 176 AD cases and 187 controls. Our results confirm the utility of BCD for identifying subtypes of heterogeneous diseases.
晚发性阿尔茨海默病(AD)是一种高度复杂的疾病,具有多种亚型,其风险因素、病理表现和临床特征各不相同。发现诊断特定阿尔茨海默病亚型的生物标志物是了解这种神秘疾病的生物机制、产生候选药物靶点和选择药物试验参与者的关键一步。评估候选生物标记物的常用统计方法--折叠变化(FC)和接收者工作特征曲线下面积(AUC)--是针对同质数据设计的,我们证明了这些方法在用于评估占患病病例不到一半的亚型时存在固有的缺陷。我们引入了一种独特的评估指标,它基于值的分布而不是值的大小,以确定与患病病例子集相关的分析物,从而揭示亚型的潜在生物标记物。我们的方法--双峰系数差(BCD)--计算病例和对照组的双峰程度之差。我们用包含近乎完美亚型的大规模合成数据试验证明了我们方法的有效性。为了揭示异质性亚型的新型 AD 生物标记物,我们将 BCD 应用于 176 例 AD 病例和 187 例对照的 8650 个基因的基因表达数据。我们的结果证实了 BCD 在识别异质性疾病亚型方面的实用性。
{"title":"Capturing biomarkers associated with Alzheimer disease subtypes using data distribution characteristics","authors":"Kenneth Smith, Sharlee Climer","doi":"10.3389/fncom.2024.1388504","DOIUrl":"https://doi.org/10.3389/fncom.2024.1388504","url":null,"abstract":"Late-onset Alzheimer disease (AD) is a highly complex disease with multiple subtypes, as demonstrated by its disparate risk factors, pathological manifestations, and clinical traits. Discovery of biomarkers to diagnose specific AD subtypes is a key step towards understanding biological mechanisms underlying this enigmatic disease, generating candidate drug targets, and selecting participants for drug trials. Popular statistical methods for evaluating candidate biomarkers, fold change (FC) and area under the receiver operating characteristic curve (AUC), were designed for homogeneous data and we demonstrate the inherent weaknesses of these approaches when used to evaluate subtypes representing less than half of the diseased cases. We introduce a unique evaluation metric that is based on the distribution of the values, rather than the magnitude of the values, to identify analytes that are associated with a subset of the diseased cases, thereby revealing potential biomarkers for subtypes. Our approach, Bimodality Coefficient Difference (BCD), computes the difference between the degrees of bimodality for the cases and controls. We demonstrate the effectiveness of our approach with large-scale synthetic data trials containing nearly perfect subtypes. In order to reveal novel AD biomarkers for heterogeneous subtypes, we applied BCD to gene expression data for 8,650 genes for 176 AD cases and 187 controls. Our results confirm the utility of BCD for identifying subtypes of heterogeneous diseases.","PeriodicalId":12363,"journal":{"name":"Frontiers in Computational Neuroscience","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142205487","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-30DOI: 10.3389/fncom.2024.1388224
Yue Yuan, Xiangyu Ye, Jian Cui, Junyang Zhang, Zhaoxiang Wang
IntroductionElectrical stimulation of the brain has shown promising prospects in treating various brain diseases. Although biphasic pulse stimulation remains the predominant clinical approach, there has been increasing interest in exploring alternative stimulation waveforms, such as sinusoidal stimulation, to improve the effectiveness of brain stimulation and to expand its application to a wider range of brain disorders. Despite this growing attention, the effects of sinusoidal stimulation on neurons, especially on their nonlinear firing characteristics, remains unclear.MethodsTo address the question, 50 Hz sinusoidal stimulation was applied on Schaffer collaterals of the rat hippocampal CA1 region in vivo. Single unit activity of both pyramidal cells and interneurons in the downstream CA1 region was recorded and analyzed. Two fractal indexes, namely the Fano factor and Hurst exponent, were used to evaluate changes in the long-range correlations, a manifestation of nonlinear dynamics, in spike sequences of neuronal firing.ResultsThe results demonstrate that sinusoidal electrical stimulation increased the firing rates of both pyramidal cells and interneurons, as well as altered their firing to stimulation-related patterns. Importantly, the sinusoidal stimulation increased, rather than decreased the scaling exponents of both Fano factor and Hurst exponent, indicating an increase in the long-range correlations of both pyramidal cells and interneurons.DiscussionThe results firstly reported that periodic sinusoidal stimulation without long-range correlations can increase the long-range correlations of neurons in the downstream post-synaptic area. These results provide new nonlinear mechanisms of brain sinusoidal stimulation and facilitate the development of new stimulation modes.
{"title":"Nonlinear analysis of neuronal firing modulated by sinusoidal stimulation at axons in rat hippocampus","authors":"Yue Yuan, Xiangyu Ye, Jian Cui, Junyang Zhang, Zhaoxiang Wang","doi":"10.3389/fncom.2024.1388224","DOIUrl":"https://doi.org/10.3389/fncom.2024.1388224","url":null,"abstract":"IntroductionElectrical stimulation of the brain has shown promising prospects in treating various brain diseases. Although biphasic pulse stimulation remains the predominant clinical approach, there has been increasing interest in exploring alternative stimulation waveforms, such as sinusoidal stimulation, to improve the effectiveness of brain stimulation and to expand its application to a wider range of brain disorders. Despite this growing attention, the effects of sinusoidal stimulation on neurons, especially on their nonlinear firing characteristics, remains unclear.MethodsTo address the question, 50 Hz sinusoidal stimulation was applied on Schaffer collaterals of the rat hippocampal CA1 region <jats:italic>in vivo</jats:italic>. Single unit activity of both pyramidal cells and interneurons in the downstream CA1 region was recorded and analyzed. Two fractal indexes, namely the Fano factor and Hurst exponent, were used to evaluate changes in the long-range correlations, a manifestation of nonlinear dynamics, in spike sequences of neuronal firing.ResultsThe results demonstrate that sinusoidal electrical stimulation increased the firing rates of both pyramidal cells and interneurons, as well as altered their firing to stimulation-related patterns. Importantly, the sinusoidal stimulation increased, rather than decreased the scaling exponents of both Fano factor and Hurst exponent, indicating an increase in the long-range correlations of both pyramidal cells and interneurons.DiscussionThe results firstly reported that periodic sinusoidal stimulation without long-range correlations can increase the long-range correlations of neurons in the downstream post-synaptic area. These results provide new nonlinear mechanisms of brain sinusoidal stimulation and facilitate the development of new stimulation modes.","PeriodicalId":12363,"journal":{"name":"Frontiers in Computational Neuroscience","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142205488","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-30DOI: 10.3389/fncom.2024.1422159
Manoj Kumar Nandi, Michele Valla, Matteo di Volo
Gamma oscillations (30–120 Hz) in the brain are not periodic cycles, but they typically appear in short-time windows, often called oscillatory bursts. While the origin of this bursting phenomenon is still unclear, some recent studies hypothesize its origin in the external or endogenous noise of neural networks. We demonstrate that an exact neural mass model of excitatory and inhibitory quadratic-integrate and fire-spiking neurons theoretically predicts the emergence of a different regime of intrinsic bursting gamma (IBG) oscillations without any noise source, a phenomenon due to collective chaos. This regime is indeed observed in the direct simulation of spiking neurons, characterized by highly irregular spiking activity. IBG oscillations are distinguished by higher phase-amplitude coupling to slower theta oscillations concerning noise-induced bursting oscillations, thus indicating an increased capacity for information transfer between brain regions. We demonstrate that this phenomenon is present in both globally coupled and sparse networks of spiking neurons. These results propose a new mechanism for gamma oscillatory activity, suggesting deterministic collective chaos as a good candidate for the origin of gamma bursts.
{"title":"Bursting gamma oscillations in neural mass models","authors":"Manoj Kumar Nandi, Michele Valla, Matteo di Volo","doi":"10.3389/fncom.2024.1422159","DOIUrl":"https://doi.org/10.3389/fncom.2024.1422159","url":null,"abstract":"Gamma oscillations (30–120 Hz) in the brain are not periodic cycles, but they typically appear in short-time windows, often called oscillatory bursts. While the origin of this bursting phenomenon is still unclear, some recent studies hypothesize its origin in the external or endogenous noise of neural networks. We demonstrate that an exact neural mass model of excitatory and inhibitory quadratic-integrate and fire-spiking neurons theoretically predicts the emergence of a different regime of intrinsic bursting gamma (IBG) oscillations without any noise source, a phenomenon due to collective chaos. This regime is indeed observed in the direct simulation of spiking neurons, characterized by highly irregular spiking activity. IBG oscillations are distinguished by higher phase-amplitude coupling to slower theta oscillations concerning noise-induced bursting oscillations, thus indicating an increased capacity for information transfer between brain regions. We demonstrate that this phenomenon is present in both globally coupled and sparse networks of spiking neurons. These results propose a new mechanism for gamma oscillatory activity, suggesting deterministic collective chaos as a good candidate for the origin of gamma bursts.","PeriodicalId":12363,"journal":{"name":"Frontiers in Computational Neuroscience","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142205489","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-29DOI: 10.3389/fncom.2024.1293279
Congzhou M. Sha, Jian Wang, Richard B. Mailman, Yang Yang, Nikolay V. Dokholyan
The question of how consciousness and behavior arise from neural activity is fundamental to understanding the brain, and to improving the diagnosis and treatment of neurological and psychiatric disorders. There is significant murine and primate literature on how behavior is related to the electrophysiological activity of the medial prefrontal cortex and its role in working memory processes such as planning and decision-making. Existing experimental designs, specifically the rodent spike train and local field potential recordings during the T-maze alternation task, have insufficient statistical power to unravel the complex processes of the prefrontal cortex. We therefore examined the theoretical limitations of such experiments, providing concrete guidelines for robust and reproducible science. To approach these theoretical limits, we applied dynamic time warping and associated statistical tests to data from neuron spike trains and local field potentials. The goal was to quantify neural network synchronicity and the correlation of neuroelectrophysiology with rat behavior. The results show the statistical limitations of existing data, and the fact that making meaningful comparison between dynamic time warping with traditional Fourier and wavelet analysis is impossible until larger and cleaner datasets are available.
意识和行为是如何从神经活动中产生的,这个问题对于理解大脑、改善神经和精神疾病的诊断和治疗至关重要。关于行为如何与内侧前额叶皮层的电生理活动及其在计划和决策等工作记忆过程中的作用相关,已有大量关于小鼠和灵长类动物的文献。现有的实验设计,特别是啮齿类动物在 T 迷宫交替任务中的尖峰序列和局部场电位记录,没有足够的统计能力来揭示前额叶皮层的复杂过程。因此,我们研究了此类实验的理论限制,为稳健、可重复的科学研究提供了具体指导。为了接近这些理论限制,我们对神经元尖峰列车和局部场电位数据进行了动态时间扭曲和相关统计检验。目的是量化神经网络的同步性以及神经电生理学与大鼠行为的相关性。结果表明了现有数据在统计方面的局限性,以及在获得更大、更清晰的数据集之前,不可能将动态时间扭曲与传统的傅里叶和小波分析进行有意义的比较。
{"title":"Quantifying network behavior in the rat prefrontal cortex","authors":"Congzhou M. Sha, Jian Wang, Richard B. Mailman, Yang Yang, Nikolay V. Dokholyan","doi":"10.3389/fncom.2024.1293279","DOIUrl":"https://doi.org/10.3389/fncom.2024.1293279","url":null,"abstract":"The question of how consciousness and behavior arise from neural activity is fundamental to understanding the brain, and to improving the diagnosis and treatment of neurological and psychiatric disorders. There is significant murine and primate literature on how behavior is related to the electrophysiological activity of the medial prefrontal cortex and its role in working memory processes such as planning and decision-making. Existing experimental designs, specifically the rodent spike train and local field potential recordings during the T-maze alternation task, have insufficient statistical power to unravel the complex processes of the prefrontal cortex. We therefore examined the theoretical limitations of such experiments, providing concrete guidelines for robust and reproducible science. To approach these theoretical limits, we applied dynamic time warping and associated statistical tests to data from neuron spike trains and local field potentials. The goal was to quantify neural network synchronicity and the correlation of neuroelectrophysiology with rat behavior. The results show the statistical limitations of existing data, and the fact that making meaningful comparison between dynamic time warping with traditional Fourier and wavelet analysis is impossible until larger and cleaner datasets are available.","PeriodicalId":12363,"journal":{"name":"Frontiers in Computational Neuroscience","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142205513","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-29DOI: 10.3389/fncom.2024.1454529
Yue Hu, Jian Liu, Rencheng Sun, Yongqiang Yu, Yi Sui
IntroductionThe automatic and precise classification of epilepsy types using electroencephalogram (EEG) data promises significant advancements in diagnosing patients with epilepsy. However, the intricate interplay among multiple electrode signals in EEG data poses challenges. Recently, Graph Convolutional Neural Networks (GCN) have shown strength in analyzing EEG data due to their capability to describe complex relationships among different EEG regions. Nevertheless, several challenges remain: (1) GCN typically rely on predefined or prior graph topologies, which may not accurately reflect the complex correlations between brain regions. (2) GCN struggle to capture the long-temporal dependencies inherent in EEG signals, limiting their ability to effectively extract temporal features.MethodsTo address these challenges, we propose an innovative epileptic seizure classification model based on an Iterative Gated Graph Convolutional Network (IGGCN). For the epileptic seizure classification task, the original EEG graph structure is iteratively optimized using a multi-head attention mechanism during training, rather than relying on a static, predefined prior graph. We introduce Gated Graph Neural Networks (GGNN) to enhance the model's capacity to capture long-term dependencies in EEG series between brain regions. Additionally, Focal Loss is employed to alleviate the imbalance caused by the scarcity of epileptic EEG data.ResultsOur model was evaluated on the Temple University Hospital EEG Seizure Corpus (TUSZ) for classifying four types of epileptic seizures. The results are outstanding, achieving an average F1 score of 91.5% and an average Recall of 91.8%, showing a substantial improvement over current state-of-the-art models.DiscussionAblation experiments verified the efficacy of iterative graph optimization and gated graph convolution. The optimized graph structure significantly differs from the predefined EEG topology. Gated graph convolutions demonstrate superior performance in capturing the long-term dependencies in EEG series. Additionally, Focal Loss outperforms other commonly used loss functions in the TUSZ classification task.
导言:利用脑电图(EEG)数据对癫痫类型进行自动和精确的分类,有望在诊断癫痫患者方面取得重大进展。然而,脑电图数据中多个电极信号之间错综复杂的相互作用带来了挑战。最近,图卷积神经网络(Graph Convolutional Neural Networks,GCN)在分析脑电图数据方面显示出了优势,因为它能够描述不同脑电图区域之间的复杂关系。然而,仍然存在以下几个挑战:(1)GCN 通常依赖于预定义或先验的图拓扑结构,这可能无法准确反映大脑区域之间的复杂关联。(2) GCN 难以捕捉脑电信号固有的长时空依赖性,从而限制了其有效提取时空特征的能力。为了应对这些挑战,我们提出了一种基于迭代门控图卷积网络(IGGCN)的创新性癫痫发作分类模型。针对癫痫发作分类任务,我们在训练过程中使用多头注意机制迭代优化原始脑电图图结构,而不是依赖于静态、预定义的先验图。我们引入了门控图神经网络(GGNN),以增强模型捕捉脑区之间脑电图序列长期依赖关系的能力。此外,我们还采用了 "病灶损失"(Focal Loss)技术来缓解癫痫脑电图数据稀缺所造成的不平衡。结果非常出色,平均 F1 得分为 91.5%,平均 Recall 得分为 91.8%,与目前最先进的模型相比有了大幅提高。讨论消融实验验证了迭代图优化和门控图卷积的功效。优化后的图结构与预定义的脑电图拓扑结构有很大不同。门控图卷积在捕捉脑电图序列的长期依赖性方面表现出卓越的性能。此外,Focal Loss 在 TUSZ 分类任务中的表现优于其他常用损失函数。
{"title":"Classification of epileptic seizures in EEG data based on iterative gated graph convolution network","authors":"Yue Hu, Jian Liu, Rencheng Sun, Yongqiang Yu, Yi Sui","doi":"10.3389/fncom.2024.1454529","DOIUrl":"https://doi.org/10.3389/fncom.2024.1454529","url":null,"abstract":"IntroductionThe automatic and precise classification of epilepsy types using electroencephalogram (EEG) data promises significant advancements in diagnosing patients with epilepsy. However, the intricate interplay among multiple electrode signals in EEG data poses challenges. Recently, Graph Convolutional Neural Networks (GCN) have shown strength in analyzing EEG data due to their capability to describe complex relationships among different EEG regions. Nevertheless, several challenges remain: (1) GCN typically rely on predefined or prior graph topologies, which may not accurately reflect the complex correlations between brain regions. (2) GCN struggle to capture the long-temporal dependencies inherent in EEG signals, limiting their ability to effectively extract temporal features.MethodsTo address these challenges, we propose an innovative epileptic seizure classification model based on an Iterative Gated Graph Convolutional Network (IGGCN). For the epileptic seizure classification task, the original EEG graph structure is iteratively optimized using a multi-head attention mechanism during training, rather than relying on a static, predefined prior graph. We introduce Gated Graph Neural Networks (GGNN) to enhance the model's capacity to capture long-term dependencies in EEG series between brain regions. Additionally, Focal Loss is employed to alleviate the imbalance caused by the scarcity of epileptic EEG data.ResultsOur model was evaluated on the Temple University Hospital EEG Seizure Corpus (TUSZ) for classifying four types of epileptic seizures. The results are outstanding, achieving an average F1 score of 91.5% and an average Recall of 91.8%, showing a substantial improvement over current state-of-the-art models.DiscussionAblation experiments verified the efficacy of iterative graph optimization and gated graph convolution. The optimized graph structure significantly differs from the predefined EEG topology. Gated graph convolutions demonstrate superior performance in capturing the long-term dependencies in EEG series. Additionally, Focal Loss outperforms other commonly used loss functions in the TUSZ classification task.","PeriodicalId":12363,"journal":{"name":"Frontiers in Computational Neuroscience","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142205490","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-27DOI: 10.3389/fncom.2024.1449364
Jessica L Verpeut,Marlies Oostland
{"title":"The significance of cerebellar contributions in early-life through aging.","authors":"Jessica L Verpeut,Marlies Oostland","doi":"10.3389/fncom.2024.1449364","DOIUrl":"https://doi.org/10.3389/fncom.2024.1449364","url":null,"abstract":"","PeriodicalId":12363,"journal":{"name":"Frontiers in Computational Neuroscience","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142205491","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-08-26DOI: 10.3389/fncom.2024.1335130
Rawan El-Zghir, Natasha Gabay, Peter Robinson
A compact description of the frequency structure and topography of human alpha-band rhythms is obtained by use of the first four brain activity eigenmodes previously derived from corticothalamic neural field theory. Just two eigenmodes that overlap in frequency are found to reproduce the observed topography of the classical alpha rhythm for subjects with a single, occipitally concentrated alpha peak in their electroencephalograms. Alpha frequency splitting and relative amplitudes of double alpha peaks are explored analytically and numerically within this four-mode framework using eigenfunction expansion and perturbation methods. These effects are found to result primarily from the different eigenvalues and corticothalamic gains corresponding to the eigenmodes. Three modes with two non-overlapping frequencies suffice to reproduce the observed topography for subjects with a double alpha peak, where the appearance of a distinct second alpha peak requires an increase of the corticothalamic gain of higher eigenmodes relative to the first. Conversely, alpha blocking is inferred to be linked to a relatively small attention-dependent reduction of the gain of the relevant eigenmodes, whose effect is enhanced by the near-critical state of the brain and whose sign is consistent with inferences from neural field theory. The topographies and blocking of the mu and tau rhythms within the alpha-band are explained analogously via eigenmodes. Moreover, the observation of three rhythms in the alpha band is due to there being exactly three members of the first family of spatially nonuniform modes. These results thus provide a simple, unified description of alpha band rhythms and enable experimental observations of spectral structure and topography to be linked directly to theory and underlying physiology.
{"title":"Unified theory of alpha, mu, and tau rhythms via eigenmodes of brain activity","authors":"Rawan El-Zghir, Natasha Gabay, Peter Robinson","doi":"10.3389/fncom.2024.1335130","DOIUrl":"https://doi.org/10.3389/fncom.2024.1335130","url":null,"abstract":"A compact description of the frequency structure and topography of human alpha-band rhythms is obtained by use of the first four brain activity eigenmodes previously derived from corticothalamic neural field theory. Just two eigenmodes that overlap in frequency are found to reproduce the observed topography of the classical alpha rhythm for subjects with a single, occipitally concentrated alpha peak in their electroencephalograms. Alpha frequency splitting and relative amplitudes of double alpha peaks are explored analytically and numerically within this four-mode framework using eigenfunction expansion and perturbation methods. These effects are found to result primarily from the different eigenvalues and corticothalamic gains corresponding to the eigenmodes. Three modes with two non-overlapping frequencies suffice to reproduce the observed topography for subjects with a double alpha peak, where the appearance of a distinct second alpha peak requires an increase of the corticothalamic gain of higher eigenmodes relative to the first. Conversely, alpha blocking is inferred to be linked to a relatively small attention-dependent reduction of the gain of the relevant eigenmodes, whose effect is enhanced by the near-critical state of the brain and whose sign is consistent with inferences from neural field theory. The topographies and blocking of the mu and tau rhythms within the alpha-band are explained analogously via eigenmodes. Moreover, the observation of three rhythms in the alpha band is due to there being exactly three members of the first family of spatially nonuniform modes. These results thus provide a simple, unified description of alpha band rhythms and enable experimental observations of spectral structure and topography to be linked directly to theory and underlying physiology.","PeriodicalId":12363,"journal":{"name":"Frontiers in Computational Neuroscience","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142205514","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Prenatal alcohol exposure (PAE) refers to the exposure of the developing fetus due to alcohol consumption during pregnancy and can have life-long consequences for learning, behavior, and health. Understanding the impact of PAE on the developing brain manifests challenges due to its complex structural and functional attributes, which can be addressed by leveraging machine learning (ML) and deep learning (DL) approaches. While most ML and DL models have been tailored for adult-centric problems, this work focuses on applying DL to detect PAE in the pediatric population. This study integrates the pre-trained simple fully convolutional network (SFCN) as a transfer learning approach for extracting features and a newly trained classifier to distinguish between unexposed and PAE participants based on T1-weighted structural brain magnetic resonance (MR) scans of individuals aged 2–8 years. Among several varying dataset sizes and augmentation strategy during training, the classifier secured the highest sensitivity of 88.47% with 85.04% average accuracy on testing data when considering a balanced dataset with augmentation for both classes. Moreover, we also preliminarily performed explainability analysis using the Grad-CAM method, highlighting various brain regions such as corpus callosum, cerebellum, pons, and white matter as the most important features in the model's decision-making process. Despite the challenges of constructing DL models for pediatric populations due to the brain's rapid development, motion artifacts, and insufficient data, this work highlights the potential of transfer learning in situations where data is limited. Furthermore, this study underscores the importance of preserving a balanced dataset for fair classification and clarifying the rationale behind the model's prediction using explainability analysis.
{"title":"Deep learning for detecting prenatal alcohol exposure in pediatric brain MRI: a transfer learning approach with explainability insights","authors":"Anik Das, Kaue Duarte, Catherine Lebel, Mariana Bento","doi":"10.3389/fncom.2024.1434421","DOIUrl":"https://doi.org/10.3389/fncom.2024.1434421","url":null,"abstract":"Prenatal alcohol exposure (PAE) refers to the exposure of the developing fetus due to alcohol consumption during pregnancy and can have life-long consequences for learning, behavior, and health. Understanding the impact of PAE on the developing brain manifests challenges due to its complex structural and functional attributes, which can be addressed by leveraging machine learning (ML) and deep learning (DL) approaches. While most ML and DL models have been tailored for adult-centric problems, this work focuses on applying DL to detect PAE in the pediatric population. This study integrates the pre-trained simple fully convolutional network (SFCN) as a transfer learning approach for extracting features and a newly trained classifier to distinguish between unexposed and PAE participants based on T1-weighted structural brain magnetic resonance (MR) scans of individuals aged 2–8 years. Among several varying dataset sizes and augmentation strategy during training, the classifier secured the highest sensitivity of 88.47% with 85.04% average accuracy on testing data when considering a balanced dataset with augmentation for both classes. Moreover, we also preliminarily performed explainability analysis using the Grad-CAM method, highlighting various brain regions such as corpus callosum, cerebellum, pons, and white matter as the most important features in the model's decision-making process. Despite the challenges of constructing DL models for pediatric populations due to the brain's rapid development, motion artifacts, and insufficient data, this work highlights the potential of transfer learning in situations where data is limited. Furthermore, this study underscores the importance of preserving a balanced dataset for fair classification and clarifying the rationale behind the model's prediction using explainability analysis.","PeriodicalId":12363,"journal":{"name":"Frontiers in Computational Neuroscience","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142226292","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}