Edge optimized hybrid quantum-classical ensemble framework for EEG and MRI based epileptic seizure detection in IoMT

Area of Science:

• Neuroscience
• Quantum Computing
• Medical Imaging

Background:

• Epileptic seizure detection is crucial for clinical intervention.
• Conventional EEG analysis faces challenges like noise, variability, and high computational cost.
• Existing methods often rely on computationally intensive deep models or cloud processing.

Purpose of the Study:

• To develop a lightweight Hybrid Quantum classical Ensemble Net (HQCE-Net) for efficient seizure detection on edge devices.
• To overcome limitations of conventional EEG analysis by integrating multi-modal data (EEG and MRI).
• To enable practical and reliable neurodiagnostic monitoring in home-based and resource-limited environments.

Main Methods:

• A pipeline involving filtering, noise reduction, and feature extraction from EEG sub-band powers (δ, θ, α, β).
• Multi-modal feature fusion using Quantum Fourier Transform (QFT) and Quantum Wavelet Transform (QWT).
• Implementation on a Raspberry Pi 5 edge device with z-score normalization for fused features.

Main Results:

• HQCE-Net achieved approximately 92% accuracy and 92% performance metrics on the EEG (CHBMIT) dataset.
• On the MRI dataset, HQCE-Net achieved 98% accuracy with 98% precision/recall.
• Demonstrated strong and consistent performance across both EEG and MRI modalities on an edge device.

Conclusions:

• The proposed HQCE-Net offers a computationally efficient and accurate solution for epileptic seizure detection.
• This lightweight, hybrid quantum-classical approach is suitable for edge deployment, enhancing accessibility for neurodiagnostic monitoring.
• The study highlights the potential of quantum-classical methods for real-time medical diagnostics in diverse clinical settings.