Abstract
This study provides a simple and effective interference suppression method for the synchronous monitoring of EIT and EEG, which could be served as a reference for the synchronous monitoring of EEG and other medical electromagnetic devices.
Introduction
For patients with drug-resistant epilepsy, surgical intervention is often the most effective treatment. However, the success of such procedures relies heavily on the precise localization of the "epileptogenic zone" (EZ). While Electroencephalography (EEG) remains the gold standard for monitoring brain activity, it provides an incomplete picture. Electrical Impedance Tomography (EIT) has emerged as a promising non-invasive tool capable of mapping regional cerebral blood flow changes, offering complementary data to assist in EZ localization.
The primary challenge in simultaneous monitoring is that EIT requires the injection of high-frequency currents into the scalp. This process generates significant electrical interference that obscures EEG signals. Previous solutions involved complex hardware filters or noise templates that lacked clinical robustness. This paper proposes a software-based dual-filtering strategy to effectively remove EIT-induced noise from EEG recordings in real-time.
Methods
The researchers employed a two-stage simulation approach to validate their filtering method. First, a 3D finite element head model was constructed using COMSOL Multiphysics, incorporating distinct layers for the scalp, skull, cerebrospinal fluid (CSF), parenchyma, and ventricle tissue.
Second, clean clinical EEG data were combined with simulated EIT signals to create a realistic "noisy" dataset. Spectral analysis identified two primary types of interference:
- High-Frequency Interference: A 51.2 kHz noise stemming from the EIT drive current.
- Low-Frequency Interference (LFI): Periodic noise resulting from the rapid switching of electrode pairs (the "drive pattern").
The proposed solution applies a low-pass filter with a 100 Hz cutoff to eliminate high-frequency noise, followed by a specialized comb filter designed to target the periodic harmonics of the LFI. The method was tested across various EIT drive patterns (Polar, Adjacent, and Avis-Barber Cross) and frame rates (1 to 10 fps).
Datasets
The study utilized two primary data sources:
- CHB-MIT Scalp EEG Database: Pediatric EEG recordings from Children’s Hospital Boston, including both normal and epileptic segments.
- Simulated EIT Data: Generated via the 16-electrode 3D head model to provide controlled interference patterns.
Key Insights
The core innovation lies in recognizing the "comb" nature of EIT noise. Because the electrode switching sequence is predictable and periodic, the resulting interference appears as regularly spaced "teeth" in the frequency domain. These harmonics are integer multiples of the EIT frame rate. By knowing the EIT settings, the comb filter can be mathematically tuned to suppress these specific frequencies without affecting the underlying EEG signal. This software-based approach is hardware-independent, potentially allowing EIT and EEG systems to share the same electrodes.
Results
The dual-filtering strategy demonstrated high signal fidelity, with Pearson Correlation Coefficients (PCC) between recovered and original EEG signals consistently exceeding 0.95, and often reaching 0.99 for normal segments. Crucially, the morphology of epileptic spikes and slow waves—essential for clinical diagnosis—was preserved. The method proved robust across different drive patterns, provided the filter order was correctly matched to the EIT frame rate.
Conclusion and Future Work
This study demonstrates that a combination of low-pass and comb filtering can effectively clean EEG signals during simultaneous EIT monitoring. While the results are promising, the study's reliance on simulated EIT signals is a limitation. Future research will focus on integrating EIT and EEG into a single hardware unit and testing the algorithm in vivo to account for biological variability and non-periodic environmental noise.
Summary updated