A neuro-chip to manage brain disorders

by admin
0 comment

Newswise — Masa Shoaran’s Comprehensive Neuroengineering Laboratory Collaborating with Stephanie Lacourt in the Faculty of Engineering Soft Bioelectronic Interface Research Team Developing NeuralTree: A closed-loop neuromodulatory system-on-a-chip that can detect and alleviate disease symptoms. Thanks to a 256-channel high-resolution sensing array and an energy-efficient machine learning processor, this system can extract and classify a wide range of biomarkers from real patient data and animal models of disease. in vivothe accuracy of symptom prediction is higher.

“NeuralTree benefits from the accuracy of neural networks and the hardware efficiency of decision tree algorithms,” says Shoaran. “Binary classification tasks such as seizure and tremor detection, and multiclass tasks such as finger movement classification for neuroprosthetic applications, he was able to integrate such a complex yet energy efficient neural interface.” is the first time.”

The results will be presented at the 2022 IEEE International Solid-State Circuits Conference. It was published in the IEEE Journal of Solid State Circuitsis the flagship journal of the integrated circuit community.

Efficiency, Scalability and Versatility

NeuralTree works by extracting neural biomarkers (patterns of electrical signals known to be associated with certain neurological disorders) from brain waves. The signal is then classified to indicate, for example, whether it is a harbinger of an impending epileptic seizure or Parkinson’s tremor. When a symptom is detected, a neurostimulator, also on the chip, activates and sends out electrical pulses to block the symptom.

Shoaran explains that NeuralTree’s unique design gives the system an unprecedented level of efficiency and versatility compared to state-of-the-art systems. The chip boasts 256 input channels compared to 32 in previous machine learning embedded devices, allowing higher resolution data to be processed by the implant. The chip’s area efficient design also means that the chip is very small (3.48mm2), which offers great potential for scalability to more channels. The integration of an “energy-aware” learning algorithm that penalizes features that consume a lot of power makes NeuralTree extremely energy efficient.

In addition to these benefits, the system can detect a wider range of symptoms than other devices that have previously focused primarily on detecting epileptic seizures. The chip’s machine learning algorithm was trained on both epileptic and Parkinson’s patient datasets to accurately classify pre-recorded neural signals from both categories.

“To our knowledge, this is the first demonstration of Parkinson’s tremor detection with an on-chip classifier,” says Shoaran.

self-update algorithm

Shoaran is passionate about making neural interfaces more intelligent to enable more effective disease management and is already looking to further innovation.

“Eventually, we will be able to use neural interfaces for a wide variety of disorders, and this will require advances in algorithmic ideas and chip design. We also need collaborations with laboratories like the Soft Bioelectronic Interface Laboratory, which can develop state-of-the-art neural electrodes, and laboratories with access to high-quality patient data.”

As a next step, she is interested in enabling on-chip algorithm updates to keep up with the evolution of neural signals.

“Neural signals change, so the performance of neural interfaces degrades over time. We are constantly striving to make our algorithms more accurate and reliable, and one way to , or to enable algorithms that can self-update.”


ERC Initiation Grant 2021, funded by the Swiss National Secretariat for Education, Research and Innovation.


U. Shin, C. Ding, B. Zhu, Y. Vyza, A. Trouillet, ECM Revol, SP Lacour, M. Shoaran, “NeuralTree: A 256-Channel 0.227-μJ/Class Versatile Neural Activity Classification and Closed-Loop.” “Neuromodulation SoC” IEEE Journal of Solid State Circuits (JSSC),roll. 57, no. 11, pp. 3243-3257, Nov. 2022, doi: 10.1109/JSSC.2022.3204508.

U. Shin, L. Somappa, C. Ding, B. Zhu, Y. Vyza, A. Trouillet, SP Lacour, M. Shoaran, “A 256-Channel 0.227 μJ/Class Versatile Brain Activity Classification and Closed-Loop Neuromodulation SoC 0.004mm2-1.51μW/channel fast settling highly multiplexed mixed-signal front-end” IEEE International Solid State Circuits Conference (ISSCC)2022, Doi: 10.1109/ISSCC42614.2022.9731776.

You may also like

Leave a Comment