Multi-Modal Closed-Loop BCI Platform

Real-time system integrating EEG, ECG, and EMG for sensorimotor decision-making research. Ultra-low latency architecture validated through rigorous SIL/HIL methodologies.

Multi-Modal Closed-Loop BCI System Architecture

The closed-loop architecture: Subject → Biosignal Recording (Brain, Heart, Muscle) → Real-Time Processing → Feedback to Subject

<100ms End-to-end Latency
3 Modalities EEG + ECG + EMG
SIL/HIL Validated

Project Overview

This platform represents a comprehensive solution for real-time, closed-loop brain-computer interface research. Developed following control engineering principles, it combines multiple biosignal modalities with ultra-low-latency machine learning pipelines to enable precise investigation of sensorimotor decision-making processes.

Key Achievement

Successfully achieved end-to-end latency of <100ms for the complete processing pipeline: from biosignal acquisition through ML inference to real-time stimulus adaptation. This performance enables investigation of rapid neural processes and their interaction with sensorimotor behavior.

🧠 Multi-Modal Integration

Synchronized recording and processing of EEG, ECG, and EMG signals with precise timestamp alignment for investigating brain-heart-muscle interactions.

⚡ Real-Time ML

Optimized machine learning models with sub-50ms inference times enable online neural decoding and adaptive experimental protocols.

🔄 Closed-Loop Control

Dynamic stimulus modification based on real-time neural state, creating truly interactive brain-computer interface experiments.

System Validation Methodology

Following control engineering best practices, the platform underwent rigorous validation using both Software-in-the-Loop (SIL) and Hardware-in-the-Loop (HIL) testing before human experiments.

Software-in-the-Loop (SIL) Validation
Software-in-the-Loop (SIL): Simulated biosignals with known ground truth for algorithm validation. No participant or physical recording needed—pure software validation of the entire pipeline.
Hardware-in-the-Loop (HIL) Validation
Hardware-in-the-Loop (HIL): Physical signal generators mimicking realistic neural patterns, testing the complete hardware-software integration with actual EEG/ECG/EMG recording equipment.

✅ SIL Validation

Verified algorithm correctness, timing accuracy, and edge case handling using simulated biosignals with known ground truth.

✅ HIL Validation

Tested real hardware components with synthetic signals to ensure the physical recording chain meets latency and quality requirements.

✅ Integration Testing

End-to-end system validation under maximum data throughput and stress conditions before human subject experiments.

Timing Analysis & Performance

For real-time closed-loop BCI systems, timing is critical. We conducted comprehensive latency profiling at every stage of the processing pipeline to ensure sub-100ms end-to-end performance.

Timing and Delay Investigation Methodology
Timing Investigation Setup: Systematic measurement of delays across the entire closed-loop system, from signal acquisition to stimulus presentation. Each component's latency was measured independently and validated against system requirements.
Measured Timeline and Delays in Closed-Loop BCI
Measured System Timeline: Complete breakdown of latencies in the closed-loop BCI setup. Key measurements: t=0 (stimulus onset), t=2.39ms (signal reaches recording PC), t=52.42ms (real-time analysis complete), t≤54.42ms (analysis delay), t≤55.42ms (LSL transmission), achieving total system latency well under 100ms for real-time closed-loop operation.

Performance Metrics

Component Latency Notes
Signal Acquisition ~2.39 ms Hardware + USB transmission
Real-Time Processing ~30 ms Feature extraction + ML inference
LSL Streaming ~55 ms Network transmission delay
Stimulus Generation <10 ms Audio/visual presentation
Total End-to-End <100 ms Complete closed-loop cycle

Research Applications

This platform has enabled multiple research projects investigating the neural mechanisms of decision-making, agency, and learning.

🧭 Decoding Human Agency

Real-time EEG investigation of the "point of no return" in decision-making—the moment when actions can no longer be cancelled and agency is formed.

EEG Decision-making Agency
View Project →

❤️ Heart-Brain Interaction

ECG-EEG real-time investigation of how cardiac phase affects learning and neural processing during decision-making tasks.

ECG Learning Cardiac timing
View Project →

🎵 Prosody-on-Demand

Audio-neural interface using real-time EEG feedback to optimize speech prosody for memory enhancement in clinical populations.

EEG Memory Neural feedback
View Project →

Technical Capabilities

📡 Multi-Modal Sensors

  • EEG: Up to 64 channels, 1000 Hz sampling
  • ECG: Synchronized cardiac monitoring
  • EMG: Multi-muscle array recording
  • Lab Streaming Layer (LSL) integration

⚙️ Signal Processing

  • Real-time filtering and artifact removal
  • Multi-modal signal synchronization
  • Cardiac phase detection (R-peak)
  • Feature extraction pipelines

🧮 ML Pipeline

  • Sub-50ms inference times
  • Online learning capabilities
  • Multiple ML framework support
  • Performance monitoring

🎯 Closed-Loop Control

  • Adaptive stimulus generation
  • Timing-critical event handling
  • Multi-threaded architecture
  • <5ms timing jitter

Publications & Documentation

Research findings and technical documentation from projects using this platform.

📝 In Preparation

Detailed documentation of the system architecture, validation methodology, and experimental results from multiple research studies are currently in preparation.

🔗 Related Publications

This platform served as the technical foundation for multiple research projects. See the individual project pages for specific publications.

System architecture white paper and technical specifications coming soon

Interested in Collaboration?

This platform is available for collaborative research. If you're interested in multi-modal BCI research or need a real-time closed-loop system for your experiments, let's discuss how we can work together.

📄 View CV 💼 Connect on LinkedIn