Our research interests lie primarily in the field of design and development of mixed-mode (analog and digital) microelectronic integrated circuits and systems that interact with physiological systems. The interaction includes
- Monitoring: recording various neuro-physiological sensory information such as electrical (EEG, ECoG, etc.) and chemical signals,
- Diagnosis: Processing recorded information to decode a physiological/neurological function or dysfunction using analytical and data-driven (AI-based) algorithms,
- Treatment: Providing a feedback signal (e.g. electrical or optical stimulation) to the neural system to start/stop/modulate a particular functionality.
- Analog and Mixed Signal IC Design
- Low Power Implantable/Wearable Sensory Microsystems
- Physiological Signal Processing VLSI Implementation
- Wirelessly-Powered Medical Implants
Multi-Modal Free-Floating Neural Interface ICs for High-Coverage High-Resolution Brain Monitoring
Implantable brain-machine interfaces (BMIs) have been proven effective in monitoring, diagnosis, and treatment of various neurological disorders during past few years. The current approach in the design of these microsystems is to pack all the necessary blocks such as multi-channel recording, real-time signal processing, and multi-channel stimulation in a single chip. However, since decoding medical information of value from neurophysiological signals requires thousands of recording channels, this approach will face several scalability issues including unacceptable microchip’s physical size, concentrated heat generation that may cause tissue damage, and susceptibility of the whole system to a local failure.
In addition to scalability, the current design paradigm falls short in terms of spatial coverage. It is established that many neurological disorders can only be captured if the neural ensembles are recorded simultaneously from several sites on the brain. Additionally, for some neurological disorders, such as sleep apnea and essential tremor, recording muscle tones and respiratory signals are also necessary, which is obviously impossible for a centralized microchip connected to a maximum cm-sized electrode array.
The main objective of this project is to develop implantable neural interfaces with a new distributed multi-modal design paradigm to tackle mentioned fundamental problems with the current state of technology.
On-Silicon Programmable Brain Pattern Recognition
Decoding brain neurological activity is one of the most important and yet challenging tasks in understanding the brain functions or dysfunctions. It has led to the development of several algorithms (both analytical and data-driven) that are designed to detect a specific neurological event such as the onset of an epileptic seizure. To meet the tight space and energy requirements of a brain-implantable device, the current approach is to hard-code an application-specific algorithm in the microchip during the fabrication. However, this comes at the cost of very limited post-fabrication flexibility.
A programmable DSP (digital signal processing) unit integrated into the system allows for implementing various sophisticated algorithms on a single device, each tailored for detecting/decoding a particular neurological event, motor or cognitive function. This also opens up the possibility to update/optimize the signal processing algorithm while running the experimental measurements and adds the flexibility to use the same framework for a new application.
However, such programmable DSP demands higher power consumption that is not viable to have with the current design paradigm of the brain neural interfaces. The main objective of this project is to solve this engineering trade off without sacrificing the performance of the system as a whole.
High-Data-Rate Short-Range Low-Power Wireless Transceivers
Short-range wireless transmission is critical to implantable and wearable biomedical devices such as retinal prosthetic implants, responsive neurostimulation systems, and high throughput DNA sequencing microarrays. With the ever-increasing size of the sensor array and the improved resolution of each sensing element, modern implantable and wearable devices require more bandwidth to transmit the resulting data.
The main objective of this project is to develop a low-power short-range high-data-rate wireless transceiver to be utilized for communicating information from an implantable device to a computer, or among different units of the device.
Highly-Efficient Silicon-Integrated Inductive Power and Data Link
The vision of particle-sized brain monitoring implants is the future of brain interface microsystems. Super-miniaturized neuromonitoring implants that are powered inductively and transmits wireless data to an external reader device are the ideal instruments for the study of the brain in real-time. A localized network of such microsystems provides a valuable real-time view of neuronal activities at various sites in the brain.
Despite the availability of the microfabrication technology, the fundamental limitations in inductive power transfer and low-power signal acquisition challenge the feasibility of the ultra-miniaturized all-wireless interface. As the size of the devices gets smaller, the available power and antenna size are reduced, and so is the capability of the microsystem to transmit back the data. Issues, such as small signal to noise ratios at the microelectrodes, high amplifier noise are some of the other design bottlenecks which only get worse with further miniaturization.
The main objective of this project is the development of a highly-efficient silicon-integrated inductive power and data link that addresses some of the main design challenges mentioned above.