EPViz

EPViz visualizes predictive model outputs for scalp electroencephalography (EEG), enabling integration of PyTorch deep learning models for spatio-temporal analysis and interpretation.

Key Features:

• PyTorch model integration: Loads and applies PyTorch deep learning models to EEG data to support spatio-temporal predictive analyses.
• Overlay predictions: Overlays model predictions on original EEG time series at channel-wise or subject-level resolution.
• High-resolution export: Exports visualizations as high-resolution image files for reporting and publication.
• Spectrum visualization and statistics: Provides spectrum (frequency-domain) visualization and computes basic statistical measures of EEG data.
• Annotation editing: Permits editing of annotations within EEG datasets.
• EDF anonymization: Includes a module to anonymize EDF files for secure sharing of clinical EEG data.

Scientific Applications:

• Model development and validation: Visualize and compare predictions from PyTorch models on scalp EEG for model training, validation, and debugging.
• Spatio-temporal analysis: Analyze temporal prediction patterns across channels and subjects to study neural dynamics.
• Frequency-domain analysis: Inspect spectral features and basic statistics to support frequency-based EEG investigations.
• Clinical data sharing and privacy: Anonymize EDF files to enable privacy-preserving sharing of clinical EEG recordings.
• Annotation curation: Edit and refine event or label annotations to improve dataset quality for downstream analyses.
• Figure generation: Produce high-resolution images of model predictions and EEG traces for inclusion in manuscripts and presentations.

Methodology:

Implemented in Python; loads and applies PyTorch deep learning models to EEG; overlays predictions on EEG time series at channel-wise or subject-level; provides spectrum visualization and basic statistical computations; supports annotation editing; anonymizes EDF files; exports high-resolution images.