A-PASS

A-PASS automates processing and analysis of concurrently acquired electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) data for sleep research.

Key Features:

• Automated Processing: Eliminates manual intervention by automating preprocessing and analysis of EEG-fMRI datasets.
• Deep Learning Integration: Uses a deep learning model trained on a sleep EEG-fMRI dataset from 45 subjects to perform sleep stage scoring from EEG acquired during fMRI.
• fMRI Indices Calculation: Computes various fMRI indices that characterize neurophysiological properties across different sleep stages.
• Statistical Mapping and Comparison: Generates statistical maps showing main effects of sleep stages and pairwise differences of fMRI indices across stages, with map quality comparable to manual multi-software processing.
• Model Validation and Performance: Validated on an independent dataset of 28 subjects and reported accuracy and F1-score higher than 70% for sleep stage classification.

Scientific Applications:

• Sleep Research: Enables analysis of brain activity patterns and sleep stage dynamics from simultaneous EEG-fMRI recordings.
• Neurophysiological Studies: Supports investigation of functional brain dynamics and neurophysiological maps across sleep stages using fMRI indices.

Methodology:

A deep learning model trained on EEG-fMRI data from 45 subjects performs sleep stage scoring from EEG recorded during fMRI; the pipeline computes fMRI indices per sleep stage, generates statistical maps for main and pairwise stage effects, and was validated on an independent 28-subject dataset with accuracy and F1-score >70%.