Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

E2CAR: An Efficient 2D-CNN Framework for Real-Time EEG Artifact Removal on Edge Devices

Haoliang Liu, Chengkun Cai, Xu Zhao, Lei Li

TL;DR

The paper tackles real-time EEG artifact removal on edge devices, where compute and power constraints limit using conventional 1-D CNN-based approaches. It introduces E2CAR, a 2-D CNN autoencoder with residual blocks that operates on a time-to-image representation and runs efficiently on Google's Edge TPU (Coral Dev Board Mini). The work demonstrates substantial improvements in inference time and power consumption while maintaining artifact-removal quality across EOG, EMG, and motion artifacts, validated on a Coral Dev Board Mini. This approach enables privacy-preserving, low-latency EEG processing in embedded or IoT settings and suggests broad applicability of 2-D CNN strategies for resource-constrained neural inference.

Abstract

Electroencephalography (EEG) signals are frequently contaminated by artifacts, affecting the accuracy of subsequent analysis. Traditional artifact removal methods are often computationally expensive and inefficient for real-time applications in edge devices. This paper presents a method to reduce the computational cost of most existing convolutional neural networks (CNN) by replacing one-dimensional (1-D) CNNs with two-dimensional (2-D) CNNs and deploys them on Edge Tensor Processing Unit (TPU), which is an open-resource hardware accelerator widely used in edge devices for low-latency, low-power operation. A new Efficient 2D-CNN Artifact Removal (E2CAR) framework is also represented using the method above, and it achieves a 90\% reduction in inference time on the TPU and decreases power consumption by 18.98\%, while maintaining comparable artifact removal performance to existing methods. This approach facilitates efficient EEG signal processing on edge devices.

E2CAR: An Efficient 2D-CNN Framework for Real-Time EEG Artifact Removal on Edge Devices

TL;DR

The paper tackles real-time EEG artifact removal on edge devices, where compute and power constraints limit using conventional 1-D CNN-based approaches. It introduces E2CAR, a 2-D CNN autoencoder with residual blocks that operates on a time-to-image representation and runs efficiently on Google's Edge TPU (Coral Dev Board Mini). The work demonstrates substantial improvements in inference time and power consumption while maintaining artifact-removal quality across EOG, EMG, and motion artifacts, validated on a Coral Dev Board Mini. This approach enables privacy-preserving, low-latency EEG processing in embedded or IoT settings and suggests broad applicability of 2-D CNN strategies for resource-constrained neural inference.

Abstract

Electroencephalography (EEG) signals are frequently contaminated by artifacts, affecting the accuracy of subsequent analysis. Traditional artifact removal methods are often computationally expensive and inefficient for real-time applications in edge devices. This paper presents a method to reduce the computational cost of most existing convolutional neural networks (CNN) by replacing one-dimensional (1-D) CNNs with two-dimensional (2-D) CNNs and deploys them on Edge Tensor Processing Unit (TPU), which is an open-resource hardware accelerator widely used in edge devices for low-latency, low-power operation. A new Efficient 2D-CNN Artifact Removal (E2CAR) framework is also represented using the method above, and it achieves a 90\% reduction in inference time on the TPU and decreases power consumption by 18.98\%, while maintaining comparable artifact removal performance to existing methods. This approach facilitates efficient EEG signal processing on edge devices.
Paper Structure (25 sections, 4 equations, 2 figures, 5 tables)

This paper contains 25 sections, 4 equations, 2 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. EEG Artifact Removal
  4. Embedded Application for EEG
  5. Methodology
  6. Data Preparation and Pre-processing
  7. Model Structure
  8. Model Optimization
  9. Evaluation Metrics
  10. Relative Root-Mean-Square Error
  11. Correlation Coefficient
  12. Computational Time and Power Consumption
  13. Experiment
  14. Model Implementation
  15. Coral Dev Board Mini Deployment
  16. ...and 10 more sections

Figures (2)

  • Figure 1: E2CAR output performance on Coral Dev board mini
  • Figure 2: Overview of the proposed E2CAR framework, where the 2D-CNN autoencoder with residual connections processes EEG signals for efficient feature extraction and artifact suppression while preserving essential signal characteristics.