Bi-Band ECoGNet for ECoG Decoding on Classification Task

TL;DR

This work introduces Bi-Band ECoGNet, a lightweight deep network for multiclass ECoG decoding that replaces MST-based temporal-frequency analysis with a Bi-BCWT module combining low- and high-frequency information. It adds a 2D Spatial-Temporal feature encoder to leverage the 8×16 electrode grid, and fuses features via an EEGNet-inspired head to achieve robust classification on visual ECoG data from macaques. The method achieves higher accuracy, smaller model size, and much faster training than prior MST-based approaches, with notable improvements on subject-specific data and insightful ablations on frequency content and spatial encoding. These results underscore the value of jointly modeling frequency-domain features and 2D spatial structure for efficient, real-time ECoG decoding in BCI applications.

Abstract

In the application of brain-computer interface (BCI), being able to accurately decode brain signals is a critical task. For the multi-class classification task of brain signal ECoG, how to improve the classification accuracy is one of the current research hotspots. ECoG acquisition uses a high-density electrode array and a high sampling frequency, which makes ECoG data have a certain high similarity and data redundancy in the temporal domain, and also unique spatial pattern in spatial domain. How to effectively extract features is both exciting and challenging. Previous work found that visual-related ECoG can carry visual information via frequency and spatial domain. Based on this finding, we focused on using deep learning to design frequency and spatial feature extraction modules, and proposed a Bi-Band ECoGNet model based on deep learning. The main contributions of this paper are: 1) The Bi-BCWT (Bi-Band Channel-Wise Transform) neural network module is designed to replace the time-consume method MST, this module greatly improves the model calculation and data storage efficiency, and effectively increases the training speed; 2) The Bi-BCWT module can effectively take into account the information both in low-frequency and high-frequency domain, which is more conducive to ECoG multi-classification tasks; 3) ECoG is acquired using 2D electrode array, the newly designed 2D Spatial-Temporal feature encoder can extract the 2D spatial feature better. Experiments have shown that the unique 2D spatial data structure can effectively improve classification accuracy; 3) Compared with previous work, the Bi-Band ECoGNet model is smaller and has higher performance, with an accuracy increase of 1.24%, and the model training speed is increased by 6 times, which is more suitable for BCI applications.

Figures (10)

• Figure 1: Outline of Data Flow of Bi-Band ECoGNet
• Figure 2: Structure of Bi-Band Channel-Wise Transform Each blue rectangle represent one TCN, there are 2 different size of TCNs. short for length of 32, long for length of 512. Each TCN will generate one Spatial-Temporal Feature map
• Figure 3: Structure of Spatial-Temporal Feature Encoder Each feature map will be reshaped to 3D structure, then processed by 2 layer of 2D kernels, finally output of feature vector.
• Figure 4: Outline of ECoG Record Each image will be used as visual stimuli, last 300 ms, ECoG signal will be measured via electrode array.
• Figure 5: Performance of Different Kernel Number Results show that the more kernel (TCN), the better performance.