Scholarlink

When modelling different types of brain data that vary in both space and time, relying on a single modality of data only offers a partial insight into the complex patterns of brain activity. Therefore, this study proposes the integration of multimodal spatiotemporal brain data (STBD) to enable a more comprehensive exploration of neuronal dynamics, drawing from various data collection methods. This approach leads to an enhanced accuracy for brain data modelling including classification and pattern recognition. The foundation of this methodology is built upon a brain-inspired spiking neural network (SNN) architecture known as NeuCube. In this research, two distinct sets of brain data, Electroencephalogram (EEG) and functional Magnetic Resonance Imaging (fMRI), are employed to train and test the SNN models through a data integration strategy. Notably, this research achieved a significant 8% improvement in classification accuracy by employing an integrated EEG-fMRI SNN model when compared to single modalities such as EEG-SNN or fMRI-SNN. The utilization of spiking neural networks, such as NeuCube, provides a robust and biologically inspired framework for modelling integrated EEG-fMRI data. Unlike traditional methods that often analyze EEG and fMRI data separately or fuse them at the feature level, this study proposes a unified SNN model that simultaneously processes both modalities at the same timepoints. By aligning EEG and fMRI data temporally, the model captures synchronized spatiotemporal dynamics, which is crucial for understanding the brain’s complex functions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Integrating Multimodal Spatiotemporal Brain Data with Spiking Neural Networks

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Maryam Doborjeh,
  • Zien Huang,
  • Zohreh Doborjeh

摘要

When modelling different types of brain data that vary in both space and time, relying on a single modality of data only offers a partial insight into the complex patterns of brain activity. Therefore, this study proposes the integration of multimodal spatiotemporal brain data (STBD) to enable a more comprehensive exploration of neuronal dynamics, drawing from various data collection methods. This approach leads to an enhanced accuracy for brain data modelling including classification and pattern recognition. The foundation of this methodology is built upon a brain-inspired spiking neural network (SNN) architecture known as NeuCube. In this research, two distinct sets of brain data, Electroencephalogram (EEG) and functional Magnetic Resonance Imaging (fMRI), are employed to train and test the SNN models through a data integration strategy. Notably, this research achieved a significant 8% improvement in classification accuracy by employing an integrated EEG-fMRI SNN model when compared to single modalities such as EEG-SNN or fMRI-SNN. The utilization of spiking neural networks, such as NeuCube, provides a robust and biologically inspired framework for modelling integrated EEG-fMRI data. Unlike traditional methods that often analyze EEG and fMRI data separately or fuse them at the feature level, this study proposes a unified SNN model that simultaneously processes both modalities at the same timepoints. By aligning EEG and fMRI data temporally, the model captures synchronized spatiotemporal dynamics, which is crucial for understanding the brain’s complex functions.