DSTA-Net:用于运动图像分类的动态时空特征增强网络
Liang Chang1, Banghua Yang1, Jiayang Zhang2
1School of Mechatronic Engineering and Automation, Research Center of Brain-Computer Engineering, Shanghai University, Shanghai, 200444 China.
Cognitive neurodynamics
|July 28, 2025
概括
动态空间时间特征增强网络 (DSTA-Net) 通过增强从EEG数据中的空间时间特征提取来提高中风康复的运动图像解码精度.
科学领域:
- 神经科学是一个神经科学.
- 生物医学工程 生物医学工程
- 机器学习 机器学习
背景情况:
- 精确解码运动图像 (MI) 对于在中风康复中推进MI应用至关重要.
- 在MI-脑电图 (EEG) 中,非静态性和高 intra-class 变异性对可靠的时空特征提取构成挑战.
- 现有的方法很难有效地捕捉MI-EEG信号的复杂动态.
研究的目的:
- 开发一个新的网络,DSTA-Net,用于增强机动图像 (MI) 解码.
- 提高MI-EEG信号分析用于中风康复的准确性和可解释性.
- 解决MI-EEG数据中非静止性和高 intra-class 变异性的挑战.
主要方法:
- 提出了集成DSTA和时空卷积 (STC) 模块的动态时空特征增强网络 (DSTA-Net).
- DSTA模块用于α和β频段的多尺度时间卷积内核和多层空间特征的分组空间卷积.
- STC模块进一步提取了用于分类的特征,DeepLIFT,共同空间模式和t-SNE用于可解释性分析.
主要成果:
- 在多个公开和自主收集的中风数据集中,DSTA-Net显示了比ShallowConvNet显著的准确性改进 (例如,BCI-IV-2a上的6.29%,OpenBMI上的3.99%).
- 在十倍交叉验证中,获得的平均精度增长为6.29% (p < 0.01),3.05% (p < 0.01),5.26% (p < 0.01),和2.25%.
- 解释性分析证实了该模型能够识别关键的EEG通道和空间模式.
结论:
- DSTA-Net有效地增强了时空特征提取,以提高机动图像 (MI) 解码精度.
- 该网络的优势为基于MI的中风康复研究和应用提供了有希望的新见解.
- 开发的模型为分析复杂的MI-EEG信号提供了一个强大的框架.
相关概念视频
01:20Motor Unit Stimulation
2.0K
When the neuron of a motor unit fires an action potential, it triggers a series of events, leading to a twitch contraction in the muscle fibers. The process of excitation-contraction coupling is crucial in relaying the action potential to the muscle fibers.
The latent period of contraction marks the onset of excitation-contraction coupling, when the action potential propagates across the sarcolemma, preparing the muscle fibers for contraction. As the fibers enter the contraction phase, the...
The latent period of contraction marks the onset of excitation-contraction coupling, when the action potential propagates across the sarcolemma, preparing the muscle fibers for contraction. As the fibers enter the contraction phase, the...
2.0K
01:14Motor and Sensory Areas of the Cortex
4.6K
The cerebral cortex, the brain's outermost layer, is pivotal in processing complex cognitive tasks, emotions, and various sensory inputs and executing voluntary motor activities. This intricate structure is divided into three primary functional areas: the motor areas, sensory areas, and association areas.
Motor Areas
The motor areas located in the frontal lobe are central to controlling voluntary movements. This region is further subdivided into the primary motor cortex and the premotor cortex....
Motor Areas
The motor areas located in the frontal lobe are central to controlling voluntary movements. This region is further subdivided into the primary motor cortex and the premotor cortex....
4.6K
01:24Sequence Networks of Rotating Machines
142
A Y-connected synchronous generator, grounded through a neutral impedance, is designed to produce balanced internal phase voltages with only positive-sequence components. The generator's sequence networks include a source voltage that is exclusively in the positive-sequence network. The sequence components of line-to-ground voltages at the generator terminals illustrate this configuration.
Zero-sequence current induces a voltage drop across the generator's neutral impedance and other...
Zero-sequence current induces a voltage drop across the generator's neutral impedance and other...
142