Design of a Closed-Loop, Bidirectional Brain Machine Interface System With Energy Efficient Neural Feature Extraction...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Electrical Engineering

Background:

• Closed-loop brain-machine interfaces (BMIs) are crucial for understanding neural circuits and developing advanced neuroprosthetics.
• Existing systems often face limitations in power consumption, miniaturization, and real-time processing capabilities for freely behaving subjects.

Purpose of the Study:

• To develop and characterize a novel, integrated, low-power bidirectional BMI microsystem on a chip (SoC).
• To enable sophisticated closed-loop neuroscience research, particularly in freely behaving animal models.

Main Methods:

• Fabrication of a 16-channel system-on-chip (SoC) using 0.18 μm CMOS technology.
• Integration of ultra-low-power neural recording front-ends, neural feature extraction units (including energy extraction and action potential detection), and programmable neural stimulator back-ends.
• Implementation of in-channel programmable PID controllers and dual analog-to-digital converters (ADCs) for neural signal digitization.
• Incorporation of a multi-mode stimulator and a low-power microcontroller with Bluetooth for wireless communication and system configuration.

Main Results:

• The fabricated SoC occupies a small silicon area (3.7 mm²) and exhibits low power dissipation (56 μW/ch).
• The system supports bidirectional communication, enabling simultaneous neural recording and stimulation with precise closed-loop control.
• The neural feature extraction units demonstrate efficient processing, including logarithmic domain frequency tuning and current-mode action potential detection.
• The stimulator offers flexible configuration for various stimulation paradigms.

Conclusions:

• The developed bidirectional BMI microsystem offers a powerful, integrated, and low-power solution for advanced neuroscience research.
• The proposed circuit techniques and system topology are applicable to a broad range of neurophysiology studies, particularly those requiring closed-loop control in mobile subjects.