Yuanyi Ding

Objective. Accurate detection and classification of high-frequency oscillations (HFOs) in electroencephalography (EEG) recordings have become increasingly important for identifying epileptogenic zones in patients with drug-resistant epilepsy. However, few open-source platforms offer both state-of-the-art computational methods and user-friendly interfaces to support practical clinical use. Approach. We present PyHFO 2.0, an enhanced open-source, Python-based platform that extends previous work by incorporating a more comprehensive set of detection methods and deep learning (DL) tools for HFO analysis. The platform now supports three commonly used detectors: short-term energy, Montreal Neurological Institute, and a newly integrated Hilbert transform-based detector. For HFO classification, PyHFO 2.0 includes DL models for artifact rejection, spike HFO detection, and identification of epileptogenic HFOs. These models are integrated with the Hugging Face ecosystem for automatic loading and can be replaced with custom-trained alternatives. An interactive annotation module enables clinicians and researchers to inspect, verify, and reclassify events. Main results. All detection and classification modules were evaluated using clinical EEG datasets, supporting the applicability of the platform in both research and translational settings. Validation across multiple datasets demonstrated close alignment with expert-labeled annotations and standard tools such as RIPPLELAB. Significance. PyHFO 2.0 aims to simplify the use of computational neuroscience tools in both research and clinical environments by combining methodological rigor with a user-friendly graphical interface. Its scalable architecture and model integration capabilities support a range of applications in biomarker discovery, epilepsy diagnostics, and clinical decision support, bridging advanced computation and practical usability.

The ability to retrieve a single episode encountered just once is a hallmark of human intelligence and episodic memory[1]. Yet, decoding a specific memory from neuronal activity in the human brain remains a formidable challenge. Here, we develop a transformer neural network model[2, 3] trained on neuronal spikes from intracranial microelectrodes recorded during a single viewing of an audiovisual episode. Combining spikes throughout the brain via cross-channel attention[4], capable of discovering neural patterns spread across brain regions and timescales, individual participant models predict memory retrieval of specific concepts such as persons or places. Brain regions differentially contribute to memory decoding before and after sleep. Models trained using only medial temporal lobe (MTL) spikes significantly decode concepts before but not after sleep, while models trained using only frontal cortex (FC) spikes decode concepts after but not before sleep. These findings suggest a system-wide distribution of information across neural populations that transforms over wake/sleep cycles[5]. Such decoding of internally generated memories suggests a path towards brain-computer interfaces to treat episodic memory disorders through enhancement or muting of specific memories.

Accurate detection and classification of high-frequency oscillations (HFOs) in electroencephalography (EEG) recordings have become increasingly important for identifying epileptogenic zones in patients with drug-resistant epilepsy. However, few open-source platforms offer comprehensive and accessible tools that integrate conventional signal processing with modern deep learning approaches for biomarker analysis. We introduce PyHFO 2.0, an enhanced platform designed for automated detection, classification, and expert annotation of neural events. PyHFO 2.0 includes three commonly used detection methods: short-term energy (STE), the Montreal Neurological Institute (MNI) approach, and a Hilbert transform-based detector. For classification, the platform incorporates deep learning models for artifact rejection, spike-associated HFO (spkHFO) detection, and epileptogenic HFO (eHFO) identification. These models are integrated with the Hugging Face ecosystem for seamless loading and can be replaced with custom-trained alternatives. Furthermore, PyHFO 2.0 features an interactive annotation interface that enables clinicians and researchers to inspect, verify, and refine automated results. The platform was validated using clinical EEG datasets from both human and rodent models of epilepsy, confirming its reliability. PyHFO 2.0 aims to simplify the use of computational neuroscience tools in both research and clinical environments by combining methodological rigor with a user-friendly graphical interface. Its scalable architecture and model integration capabilities support a range of applications in biomarker discovery, epilepsy diagnostics, and clinical decision support, bridging advanced computation and practical usability.

Objective. This study aims to develop and validate an end-to-end software platform, PyHFO, that streamlines the application of deep learning (DL) methodologies in detecting neurophysiological biomarkers for epileptogenic zones from EEG recordings. Approach. We introduced PyHFO, which enables time-efficient high-frequency oscillation (HFO) detection algorithms like short-term energy and Montreal Neurological Institute and Hospital detectors. It incorporates DL models for artifact and HFO with spike classification, designed to operate efficiently on standard computer hardware. Main results. The validation of PyHFO was conducted on three separate datasets: the first comprised solely of grid/strip electrodes, the second a combination of grid/strip and depth electrodes, and the third derived from rodent studies, which sampled the neocortex and hippocampus using depth electrodes. PyHFO demonstrated an ability to handle datasets efficiently, with optimization techniques enabling it to achieve speeds up to 50 times faster than traditional HFO detection applications. Users have the flexibility to employ our pre-trained DL model or use their EEG data for custom model training. Significance. PyHFO successfully bridges the computational challenge faced in applying DL techniques to EEG data analysis in epilepsy studies, presenting a feasible solution for both clinical and research settings. By offering a user-friendly and computationally efficient platform, PyHFO paves the way for broader adoption of advanced EEG data analysis tools in clinical practice and fosters potential for large-scale research collaborations.

In the context of epilepsy studies, intracranially-recorded interictal high-frequency oscillations (HFOs) in EEG signals are emerging as promising spatial neurophysiological biomarkers for epileptogenic zones. While significant efforts have been made in identifying and understanding these biomarkers, deep learning is carving novel avenues for biomarker detection and analysis. Yet, transitioning such methodologies to clinical environments is difficult due to the rigorous computational needs of processing EEG data via deep learning. This paper presents our development of an advanced end to end software platform, PyHFO, aimed at bridging this gap. PyHFO provides an integrated and user-friendly platform that includes time-efficient HFO detection algorithms such as short-term energy (STE) and Montreal Neurological Institute and Hospital (MNI) detectors and deep learning models for artifact and HFO with spike classification. This application functions seamlessly on conventional computer hardware. Our platform has been validated to adeptly handle datasets from 10-minute EEG recordings captured via grid/strip electrodes in 19 patients. Through implementation optimization, PyHFO achieves speeds up to 50 times faster than the standard HFO detection method. Users can either employ our pre-trained deep learning model for their analyses or use their EEG data to train their model. As such, PyHFO holds great promise for facilitating the use of advanced EEG data analysis tools in clinical practice and large-scale research collaborations.