Get email delivery of the Cadence blog featured here
Smart, connected medical implants are potentially among the most worthwhile of Internet of Things (IoT) devices. (Who really needs a smart water fountain for their cat?) The opportunity to improve or even save lives is one of the things that drives UC Berkeley Assistant Professor Rikky Muller. Muller co-founded Cortera Neurotechnologies, which designs medical devices that strive to improve patient care and quality of life and to advance neuroscientific research. She discussed her experiences in a SKY Talk at the DAC Pavilion on June 6 during this year's Design Automation Conference in Austin.
“Recent advances in brain/machine interfaces have offered hope to 100 million people worldwide who are living with paralysis,” Muller noted. However, “even state-of-the-art solutions clearly have a long way to go.”
Devices used for various treatments tend to be big and bulky, wired, and generally uncomfortable and/or inconvenient for the patient. Small, minimally invasive wireless neural interfaces offer promise, says Muller, noting, “Our work aims to transition these tools into a much smaller form factor, integrating the same functionality on a chip, so it can be fully implanted and remotely powered.”
In researching and eventually developing a solution, Muller and her colleagues turned to electrocorticography (ECoG), where sensors are placed directly on the surface of the brain. Because the technology at the time was insufficient, Muller and her team made some adjustments to their prototype:
Aiming for extreme miniaturization and power efficiency, the team designed high-density ECoG electrodes that could map which region of the brain’s cortex is tuned to hearing which specific frequency. They realized that by monolithically integrating the antenna in the same MEMS process, they could afford larger diameter and up to 800mW of power, without adding to the invasiveness of the device. In the wireless IC, neural signal acquisition front-ends perform amplification, filtering, and digitization. To reduce die area and improve power efficiency, the team designed a mixed-signal architecture with a mixed-signal servo loop and a digital loop filter on a 65nm low-power CMOS process. Large time constraints were moved from the analog to the digital domain. The result? 64 channels in just 1.6mm2 and a 3X improvement in power efficiency.
Their resulting prototype, validated via testing on an anesthetized rodent, is a 64-channel, wireless, single-chip ECoG neural sensor. “We believe this technology is an excellent prospect to become a technology of choice for clinically relevant neural recording,” noted Muller, who helped form Cortera in 2013 to further develop the solution.
Muller did present “one great challenge” to the design automation community: how do we determine design specifications and how do we validate our designs? The validation process can be long and expensive. What researchers and developers need, Muller said, are multi-domain tools that incorporate models of biological systems for safety and efficacy. The result can save years of development time, millions of dollars, and, potentially, human lives.
“There’s very little doubt that we’re quickly heading toward a future full of connected devices that are going to monitor, learn, diagnose, and even treat our illnesses,” concluded Muller, who made MIT Technology Review’s list of 35 Innovators Under 35 in 2015.