Data augmentation for invasive brain-computer interfaces based on stereo-electroencephalography (SEEG)

Xiaolong Wu,Dingguo Zhang,Guangye Li,Xin Gao,Benjamin Metcalfe,Liang Chen

DOI: https://doi.org/10.1088/1741-2552/ad200e

IF: 5.043

2024-01-20

Journal of Neural Engineering

Abstract:—Objective: : Deep learning is increasingly used for Brain-computer interfaces (BCIs). However, the available brain data is sparse, especially for invasive BCIs, which can dramatically deteriorate deep learning performance. Data augmentation methods (DA), such as generative models, can help to address this issue. However, existing studies on brain signals relied on convolutional neural networks (CNNs) and ignored the temporal dependence. This paper tried to enhance the generative model by capturing the temporal relationship from a time-series perspective. Methods: A conditional generative network (cTGAN) based on the transformer model is proposed, and tested on the Stereo- electroencephalography (SEEG) data which was recorded from eight epileptic patients performing five different movements. Three other commonly-used DA methods were also implemented: noise injection (NI), variational autoencoder (VAE) and conditional Wasserstein GAN with gradient penalty (cWGANGP). Artificial SEEG data was generated, and various metrics were used to compare the data quality, including visual inspection, Cosine similarity (CS), Jensen-Shannon distance (JSD) and the effect on the performance of a deep learning-based classifier. Result: Both the proposed cTGAN and the cWGANGP methods were able to generate realistic data, while NI and VAE output inferior samples when visualised as raw sequences and in a lower dimensional space. The cTGAN generated the best samples in terms of cosine similarity and Jensen-Shannon distance and outperformed cWGANGP significantly in enhancing the performance of a deep learning-based classifier (each of them yielding a significant improvement of 6% and 3.4%, respectively). Conclusion: This paper demonstrated that a generative model that preserves temporal dependence is superior in data generation and boosting deep learning performance for SEEG signals. Significance: This is the first time that DA methods are applied to invasive BCIs based on SEEG. In addition, this study demonstrated the advantages of the model that preserves the temporal dependence from a time-series perspective.

engineering, biomedical,neurosciences

What problem does this paper attempt to address?

The problem this paper attempts to address is the scarcity of data in invasive Brain-Computer Interfaces (BCI), particularly those based on stereoelectroencephalography (SEEG). Due to the limited amount of available data, deep learning models struggle to generalize effectively to unseen instances. To alleviate this issue, the authors propose a conditional Transformer-based Generative Adversarial Network (cTGAN) to enhance the generative model and capture temporal dependencies from a time-series perspective. By generating high-quality synthetic data, the aim is to improve the performance of deep learning classifiers.

Specifically, the goals of this study include:
1. **Generating realistic SEEG data**: Creating synthetic data that resembles real data through data augmentation methods.
2. **Improving classification accuracy**: Evaluating the impact of the generated data on the performance of deep learning classifiers, particularly in a 5-class motor classification task.

To achieve these goals, the authors compare the proposed cTGAN method with three other commonly used data augmentation methods: Noise Injection (NI), Variational Autoencoder (VAE), and Conditional Wasserstein GAN with Gradient Penalty (cWGANGP). The quality of the generated data is assessed using various metrics such as visual inspection, cosine similarity, and Jensen-Shannon distance, and the effectiveness of different methods is measured by changes in classification accuracy.