Decoding motor imagery based on electroencephalography (EEG) is limited by high data noise and high model computational complexity. Starting from EEGNet, this study achieves high-accuracy decoding through three steps. First, frequency domain analysis was performed to reveal the frequency modeling patterns of deep learning models. Utilizing prior knowledge from brain science regarding the key frequency bands for motor imagery, we adjusted the convolution kernels and pooling sizes of EEGNet to focus on effective frequency bands. Subsequently, a residual network was introduced to preserve high-frequency detailed features. Finally, temporal convolution modules were used to deeply capture temporal dependencies, significantly enhancing feature discriminability. Experiments were conducted on the BCI Competition IV 2a and 2b datasets. The 2a dataset includes multi-channel data with 22 channels, while the 2b dataset contains low-channel data with only 3 channels, reflecting significant scenario differences. Our method achieved average classification accuracies of 86.23% and 86.75% respectively, surpassing advanced models like EEG-Conformer and EEG-TransNet. Meanwhile, the Multiply–accumulate operations (MACs) were 27.16 M, a reduction of over 50% compared to the comparison models, and the Forward/Backward Pass Size was 14.33 MB. This significantly reduced computational complexity and memory footprint. This paper employs the simplest and most fundamental techniques in its design, highlighting the critical role of brain science knowledge in model development. The proposed method demonstrates broad application potential.

Objective: The transmission and storage of large-scale EEG data require high-ratio EEG compression. However, existing EEG compression methods struggle to achieve high compression efficiency while preserving reconstruction quality due to statistical redundancy and the loss of high-frequency information at extreme compression ratios. Methods: To address these limitations, we propose TFANet, a novel high-ratio EEG compression framework that integrates autoencoder learning with entropy coding to optimize the latent space distribution, effectively reducing redundancy and maximizing compression efficiency. To address the issue of high-frequency information loss in existing methods, which leads to significant detail degradation at high compression ratios, we propose the frequency attention block (FAB) and the time-frequency enhancement block (TFEB). FAB leverages fast fourier transform for global frequency-aware compression, while TFEB integrates discrete wavelet transform with channel attention to preserve fine-grained time-frequency features. By utilizing global frequency awareness to guide local feature extraction, our approach ensures more effective retention of critical EEG details. Results: Experiments on public EEG datasets show that TFANet achieves an unprecedented 333× compression ratio while maintaining superior reconstruction quality, significantly outperforming existing methods. Conclusion: These results highlight TFANet's potential for large-scale EEG applications, enabling efficient data transmission and storage while preserving critical neural information. Significance: TFANet reduces the storage and transmission costs of large-scale EEG data, laying the foundation for its practical applications in medical diagnosis and remote monitoring.