PyHFO 2.0: an open-source platform for deep learning-based clinical high-frequency oscillations analysis.

Area of Science:

• Computational Neuroscience
• Medical Informatics
• Epileptology

Background:

• Accurate detection and classification of high-frequency oscillations (HFOs) in electroencephalography (EEG) are crucial for identifying epileptogenic zones in drug-resistant epilepsy.
• Existing open-source platforms often lack comprehensive computational methods and user-friendly interfaces for practical clinical application.

Purpose of the Study:

• To introduce PyHFO 2.0, an enhanced open-source Python platform for HFO analysis.
• To integrate advanced detection algorithms and deep learning models for HFO classification, artifact rejection, and epileptogenic zone identification.

Main Methods:

• PyHFO 2.0 incorporates three HFO detectors: short-term energy, Montreal Neurological Institute, and Hilbert transform-based.
• Deep learning models for artifact rejection and HFO classification are integrated via the Hugging Face ecosystem.
• An interactive annotation module allows for manual inspection, verification, and reclassification of detected events.

Main Results:

• All detection and classification modules were validated on clinical EEG datasets, demonstrating applicability in research and translational settings.
• Performance showed close alignment with expert annotations and established tools like RIPPLELAB across multiple datasets.

Conclusions:

• PyHFO 2.0 simplifies the clinical and research use of computational neuroscience tools through a user-friendly interface and methodological rigor.
• The platform's architecture supports biomarker discovery, epilepsy diagnostics, and clinical decision support by bridging advanced computation and practical usability.