基于频道注意力和时空平行处理的轻量级EEG阶段预测
Shufei Duan1,2, Yuting Yan2, Qianrong Guo2
1College of Computer Science and Technology, Shanxi University of Electronic Science and Technology, Linfen 041000, China.
Brain sciences
|January 28, 2026
概括
新的深度学习模型改善了实时电脑图 (EEG) 阶段预测,用于闭环,相锁跨磁刺激 (TMS). 这减少了时间错误,提高了刺激精度和一致性,以获得更好的治疗结果.
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:
- 闭环相锁跨磁刺激 (TMS) 需要精确的实时电脑电图 (EEG) 阶段预测.
- 计时不准确性,特别是在EEG信号高峰和低谷附近,显著损害了准准确性.
- 现有的预测方法在实现有效的闭环控制所需的低延迟和高一致性方面面临挑战.
研究的目的:
- 为了对经典和反复的神经网络 (RNN) 预测器进行对比,用于EEG阶段预测.
- 开发新的深度学习模型,增强相位预测的一致性并减少时间延迟,特别是在信号极端的情况下.
- 引入一个新的指标,即平均滞后时间 (MLT),用于评估极端特定预测性能.
主要方法:
- 在莫纳什大学TEPs-MEPs数据集上对AR,FFT,LSTM和GRU预测者的基准测试.
- 提出一个平行DSC-Attention-GRU架构,用于高效的时空特征提取和基于注意力的依赖性建模.
- 开发用于实时应用的轻量级SqueezeNet-Attention-GRU变体,并使用MLT,PLV,APE,MAE和RMSE评估性能.
主要成果:
- LSTM和GRU模型显示,与AR/FFT相比,时间动态有所改善,但保留了剩余滞后.
- 拟议的DSC-Attention-GRU模型始终提高了相位预测的准确性,并减少了极端滞后 (MLT从7.77-7.79毫秒减少到7.50-7.56毫秒).
- 轻量级变体实现了3.7%的推断加快速度,同时保持了稳定的性能.
结论:
- 使用MLT明确优化极端定时对于闭环TMS至关重要.
- 整合深度可分离卷积 (DSC) 和注意力机制可以增强多通道建模,以减少峰值/深度滞后.
- 开发的模型提供了改进的相一致预测,支持低延迟闭环相锁定TMS的进步.
更多相关视频
09:37Measurement of Neurophysiological Signals of Ignoring and Attending Processes in Attention Control
Published on: July 5, 2015
9.5K
08:45Mapping Cortical Dynamics Using Simultaneous MEG/EEG and Anatomically-constrained Minimum-norm Estimates: an Auditory Attention Example
Published on: October 24, 2012
15.2K
相关概念视频
01:20Parallel Processing
703
The brain processes sensory information rapidly due to parallel processing, which involves sending data across multiple neural pathways at the same time. This method allows the brain to manage various sensory qualities, such as shapes, colors, movements, and locations, all concurrently. For instance, when observing a forest landscape, the brain simultaneously processes the movement of leaves, the shapes of trees, the depth between them, and the various shades of green. This enables a quick and...
703
01:23Parallel Resonance
551
The parallel RLC circuit is an arrangement where the resistor (R), inductor (L), and capacitor (C) are all connected to the same nodes and, as a result, share the same voltage across them. The parallel RLC circuit is analyzed in terms of admittance (Y), which reflects the ease with which current can flow. The admittance is given by:
551
02:27Predicting Molecular Geometry
45.7K
VSEPR Theory for Determination of Electron Pair Geometries
45.7K
01:19Ion Channels
91.4K
The movement of ions like sodium, potassium, and calcium into and out of the cell is essential to maintain the electrochemical gradient in living cells. The ion channels—a class of membrane transport proteins—help maintain this ionic gradient for the smooth functioning of physiological activities such as maintaining cell size and volume, conducting nerve impulses, and gas and nutrient exchange.
Ion channels are specialized integral membrane proteins on the plasma membrane that allow...
Ion channels are specialized integral membrane proteins on the plasma membrane that allow...
91.4K
02:31Phase Transitions
23.1K
Whether solid, liquid, or gas, a substance's state depends on the order and arrangement of its particles (atoms, molecules, or ions). Particles in the solid pack closely together, generally in a pattern. The particles vibrate about their fixed positions but do not move or squeeze past their neighbors. In liquids, although the particles are closely spaced, they are randomly arranged. The position of the particles are not fixed—that is, they are free to move past their neighbors to...
23.1K
01:23Resistors In Parallel
6.1K
Resistors are in parallel when one end of all the resistors are connected to a continuous wire of negligible resistance and the other end of all the resistors are also connected to one another through a continuous wire of negligible resistance. In the case of a parallel configuration, the potential drop across each resistor is the same. Current through each resistor can be found using Ohm’s law, I = V/R, where the voltage is constant across each resistor. The sum of the individual currents...
6.1K