Personalizing Seizure Detection for Individual Patients by Optimal Selection of EEG Signals DOI Creative Commons
Rosanna Ferrara, M Giaquinto, Gennaro Percannella

et al.

Sensors, Journal Year: 2025, Volume and Issue: 25(9), P. 2715 - 2715

Published: April 25, 2025

Electroencephalography is a widely used non-invasive method for monitoring brain electrical activity, critical diagnosing and managing neurological disorders such as epilepsy. While clinical standards use 21 electrodes to capture comprehensive neural signals, personalized approach can enhance performance by selecting patient-specific channels, reducing noise redundancy. This study introduces an innovative, lightweight deep learning system optimized real-time seizure detection in wearable devices. The uses efficient Convolutional Neural Network that processes data from just two channels. These channels are automatically selected using data-driven mechanism identifies the most informative scalp regions based on each patient's unique patterns. proposed ensures high reliability, even with small datasets, improves interpretability clinicians overcoming limitations of more complex methods. tailored channel selection boosts accuracy robust across different types while computational burden typical multi-electrode systems. Validation publicly available CHB-MIT dataset achieved average balanced 0.83 false-positive rate approximately 0.1/h. system's matches, some cases outperforms, state-of-the-art systems four fixed temporal regions, demonstrating potential two-channel solutions, specifically non-negligible 30% reduction rate. interpretable, enables development personalized, efficient, compact devices reliable everyday life.

Language: Английский

Personalizing Seizure Detection for Individual Patients by Optimal Selection of EEG Signals DOI Creative Commons
Rosanna Ferrara, M Giaquinto, Gennaro Percannella

et al.

Sensors, Journal Year: 2025, Volume and Issue: 25(9), P. 2715 - 2715

Published: April 25, 2025

Electroencephalography is a widely used non-invasive method for monitoring brain electrical activity, critical diagnosing and managing neurological disorders such as epilepsy. While clinical standards use 21 electrodes to capture comprehensive neural signals, personalized approach can enhance performance by selecting patient-specific channels, reducing noise redundancy. This study introduces an innovative, lightweight deep learning system optimized real-time seizure detection in wearable devices. The uses efficient Convolutional Neural Network that processes data from just two channels. These channels are automatically selected using data-driven mechanism identifies the most informative scalp regions based on each patient's unique patterns. proposed ensures high reliability, even with small datasets, improves interpretability clinicians overcoming limitations of more complex methods. tailored channel selection boosts accuracy robust across different types while computational burden typical multi-electrode systems. Validation publicly available CHB-MIT dataset achieved average balanced 0.83 false-positive rate approximately 0.1/h. system's matches, some cases outperforms, state-of-the-art systems four fixed temporal regions, demonstrating potential two-channel solutions, specifically non-negligible 30% reduction rate. interpretable, enables development personalized, efficient, compact devices reliable everyday life.

Language: Английский

Citations

0