Our research aims to expand the current understanding of neuronal interactions responsible for the brain’s functions/dysfunctions using intelligent microelectronic integrated circuits (ICs), and use this knowledge to devise implantable and wearable medical devices for responsive treatment of neurological disorders.

• Integrated Circuits for brain neural recording and electrical stimulation

• Implantable ElectroOptical Microsystems for high-resolution optogenetics

• Onchip Implementation of deterministic and ML-based Signal Processing algorithms

• Wireless power transfer and data communications for Implantable Devices

Accurate capture and effective control of neurological disorders that often originate in multiple regions of the brain require implantable microsystems with an ever-increasing need for larger arrays of recording and stimulation circuits. Addressing this demand while considering the energy/area constraints of brain-implantable devices necessitates a search for new circuit architectures. The focus of this aspect of our research has been on the design and development of new circuit architectures that can reduce the required silicon area and power consumption for neural recording/stimulation while maintaining all other performance metrics (e.g., signal-to-noise ratio (SNR), linearity, bandwidth, etc.) competitive to the state of the art.

Thanks to its cell type specificity, high spatiotemporal precision, and reversibility, optogenetic neuromodulation has been widely utilized in brain mapping, visual prostheses, and various neurological disorders. While a variety of optical neural interfaces have been developed, most have substantial limitations due to their size and tethering, needed to deliver either light or electricity, which may restrict the animal movements and bias the results, particularly in behavioural studies. In this aspect of our research program, we focus on the development and experimental validation of fully implantable mm-scale multichannel optogenetic closed-loop
neurostimulators capable of temperature-controlled light directivity enhanced simultaneous optical stimulation and electrical recording. We have reported various device andcircuit-level design techniques to enhance the optical stimulation light directivity, as well as the voltage compliance, linearity, and energy efficiency of the driver circuits.

Neurological events such as epileptic seizures result from interactions between large populations of neurons across the brain. Considering the uniqueness of each human’s brain and depending on the type, focus, and severity of the event, the manifestation of abnormality varies significantly from patient to patient, and over time for the same patient. Therefore, accurate detection often requires machine learning (ML)-based algorithms that are trained for each specific patient and can process a large amount of neural data recorded from many locations across the brain. The main difficulty against implementing such algorithms on an implantable device is their extremely tight power budget. The processing cannot be done in an external (non-implanted) computer either, as it requires wireless communication of raw recordings, which (a) leads to higher power consumption than on-device signal processing, and (b) causes a delay that is too long to perform timely responsive stimulation of the brain.

This aspect of our research investigates energy-efficient circuit architectures for on-chip signal processing. Capturing neurological events that build up in a matter of seconds requires detection systems with very low latency. We have investigated the design and development of various deterministic and ML-based algorithms on implantable microchips with applications in epilepsy, sleep disorders, hypertension, and Alzheimer’s disease.

Wireless powering of implantable devices allows for a significant size reduction by replacing bulky batteries with small rechargeable ones. However, for implants used for long-term applications, the device movements/rotation and the changes in energy transfer medium (i.e., body fluids) cause serious challenges in achieving a consistent acceptable power transfer efficiency from the link. For data transmission, the challenges are mostly attributed to the ever-increasing number of recording channels on a single device that demand a higher throughput for the wireless transmitter, without violating the device’s highly strict power budget set by the powering link capacity and safety measures (i.e., heat generation density).

We have integrated various far- and near-field data communication circuits, as well as impulse radio UWB transmitters in implantable neural interfaces with various data-rate (1.5-230 Mbps) and range (10mm-2m) and have investigated the trade-offs between communication and computation in these devices. For wireless power delivery, we have developed brain-implantable devices with both on-chip and off-chip coil implementations, for mm- and cm-scale devices at different implantation depths. Recently, we have been also active in the development of calibration-free ICs for link-adaptive real-time energy storage optimization of inductive power receivers.