CLoSES

CLoSES performs closed-loop intracranial electrical stimulation by detecting intracranial electroencephalography (EEG) features in real time to probe brain dynamics and evaluate electrographic biomarkers for neuromodulatory research and therapeutic applications.

Key Features:

• Real-Time Neural Feature Computation: Acquires intracranial electroencephalography (EEG), applies band-pass filtering, and continuously computes local and network neural features.
• Decision Algorithm for Stimulation Triggering: Uses a decision algorithm that triggers electrical stimulation in near real time when target neural features exceed predefined thresholds and durations.
• Flexibility and Adaptability: Supports various experimental paradigms within a single participant, including random stimulation timing, adjustable percentages of biomarker detections that lead to stimulation, and safety refractory periods to prevent overstimulation.
• Cognitive State Estimation: Implements an encoder/decoder model to estimate hidden cognitive states from neural features.
• Clinical Integration and Safety: Operates in parallel with existing clinical systems and is applied in clinical settings such as epilepsy monitoring units with implanted electrodes, incorporating safety measures for clinical deployment.

Scientific Applications:

• Human intracranial studies (12 patients): Deployed in twelve patients with depth electrodes during cognitive tasks to investigate task-related brain dynamics.
• Sleep spindle detection: Applied for detection of sleep spindles in intracranial EEG recordings.
• Epileptic activity monitoring: Used to monitor and detect epileptic activity in intracranial EEG.
• Biomarker discovery and evaluation: Facilitates discovery and evaluation of electrographic biomarkers for neurological and psychiatric disorders.
• Neuromodulatory therapy development: Enables investigation of neuromodulatory interventions and development of patient-specific therapeutic strategies.

Methodology:

Continuous acquisition and processing of intracranial EEG with band-pass filtering, continuous computation of local and network features, use of a decision algorithm to trigger stimulation based on predefined thresholds and durations, and an encoder/decoder model to estimate hidden cognitive states.