Portiloop

Portiloop enables real-time closed-loop detection of neural events from EEG and time-locked stimulation to study and modulate brain oscillations.

Key Features:

• Deep Learning Integration: Uses a fast, lightweight neural network for real-time processing of electroencephalography (EEG) to identify neural events.
• Automated Hyperparameter Optimization: Incorporates an exploration algorithm that automatically optimizes model hyperparameters to target desired brain oscillations.
• Real-Time Closed-Loop Operation: Detects neural events in real time and triggers time-locked stimulation, demonstrated on sleep spindle detection with performance comparable to offline expert analysis on the Massive Online Data Annotation (MODA) spindle dataset.
• Open-Hardware Implementation: Built from commercially available components to support portability and reduced hardware cost.

Scientific Applications:

• Fundamental neuroscience research: Enables study of neural dynamics by enabling precise, time-locked interaction with ongoing brain oscillations.
• Clinical intervention research: Supports development and testing of therapeutic stimulation strategies that modulate brain activity.
• Memory modulation studies: Applicable to experiments aiming to restore or enhance degraded memory function via targeted oscillatory stimulation (e.g., spindle-targeted interventions).

Methodology:

Portiloop captures EEG, processes signals in real time with a fast, lightweight neural network, uses an exploration algorithm for automated hyperparameter optimization to target specific oscillations, and triggers auditory, magnetic, or electrical stimuli in a closed-loop manner; validation includes sleep spindle detection assessed against the MODA dataset.