Hardware architecture for implantable brain-computer interface system-on-chip

This technology is a hardware architecture that enables real-time processing of neural data directly on an implanted brain-computer interface system-on-chip.

Unmet Need: Large-scale, safe, real-time processing of brain neural signals

Brain-computer interfaces (BCIs) are systems that connect brain neural activity to external devices, allowing for a deeper understanding of the brain. However, BCIs can also be utilized in medical assistive technologies and the development of artificial intelligence. BCIs have developed to be implantable system-on-chips (SoCs), but require large volumes of transmitted data, have limited scalability in terms of the number of neural interface sensors, and contain communication bottlenecks. There is a need for BCI SoCs that handle large-scale, safe, and real-time processing of the brain’s neural signals.

The Technology: Hardware architecture for neural SoCs to execute machine-learning models in BCIs

This technology is a hardware architecture for an implanted system-on-chip (SoC) that can execute machine-learning (ML) models in brain-computer interface (BCI) systems. The architecture enables the real-time processing of neural data directly on the implanted SoC, significantly reducing the volume of data transmitted outside the body. By doing so, the technology minimizes communication overhead and supports an increasing number of neural interface sensors, while still meeting the safety and functionality requirements of implanted devices like SoCs. By reducing the amount of external data processing, this technology has the potential to advance implantable BCI SoCs.

Applications:

Neuroscience research tool
Medical assistive technologies
Artificial intelligence development
Motion decoding
Speech synthesis
Real-time processing

Advantages:

Scalable
Energy-efficient
Reduces transmission and overhead communication
Safe
Functional

Lead Inventor:

Luca Carloni, Ph.D.

Patent Information:

Patent Pending

Related Publications:

Eichler G, Gilhotra Y, Zeng N, Kim M, Shepard K, Carloni L. 2025. “MINDFUL: Safe, Implantable, Large-Scale Brain-Computer Interfaces from a System-Level Design Perspective.” In Proceedings of the 58th IEEE/ACM International Symposium on Microarchitecture (MICRO '25). Association for Computing Machinery, New York, NY, USA. pp1672–1689.

Tech Ventures Reference:

IR CU25171
Licensing Contact: Greg Maskel