Deep EEG super-resolution: Upsampling EEG spatial resolution with Generative Adversarial Networks

2018-03-04

2018 IEEE EMBS International Conference on Biomedical & Health Informatics (BHI)

AI Summary

High-density EEG systems are prohibitively expensive, limiting spatial resolution in practical applications. To address this, we propose the first generative adversarial network (GAN)-based spatial super-resolution method for EEG, enabling end-to-end reconstruction from low-channel-count EEG signals to high-density channel configurations. Unlike conventional interpolation techniques, our approach explicitly models physiological inter-channel dependencies, jointly optimizing signal fidelity and neurophysiological plausibility. Evaluated on mental imagery task data, the method reduces MSE by approximately 104× and MAE by ~102× compared to bicubic interpolation. Critically, downstream classification accuracy remains virtually unchanged (drop <0.5%), demonstrating that reconstructed data preserve discriminative neural information. This work establishes a novel paradigm for cost-effective, high-resolution EEG acquisition without hardware modification.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Electroencephalography (EEG) activity contains a wealth of information about what is happening within the human brain. Recording more of this data has the potential to unlock endless future applications. However, the cost of EEG hardware is increasingly expensive based upon the number of EEG channels being recorded simultaneously. We combat this problem in this paper by proposing a novel deep EEG super­resolution (SR) approach based on Generative Adversarial Networks (GANs). This approach can produce high spatial resolution EEG data from low resolution samples, by generating channel-wise upsampled data to effectively interpolate numerous missing channels, thus reducing the need for expensive EEG equipment. We tested the performance using an EEG dataset from a mental imagery task. Our proposed GAN model provided ∼104 fold and ∼102 fold reduction in mean-squared error (MSE) and mean-absolute error (MAE), respectively, over the baseline bicubic interpolation method. We further validate our method by training a classifier on the original classification task, which displayed minimal loss in accuracy while using the super-resolved data. The proposed SR EEG by GAN is a promising approach to improve the spatial resolution of low density EEG headset.

Problem

• Enhance EEG spatial resolution
• Reduce EEG equipment cost
• Generate high-resolution EEG data

Innovation

• Generative Adversarial Networks enhance EEG
• Upsampling EEG spatial resolution effectively
• Reduces need for expensive EEG equipment