UC Berkeley

Neural interfaces promise to radically change medicine. Currently, amputees and persons suffering from debilitating brain disorders lack a way to regain mobility and freedom. By recording and interpreting signals from the motor control regions of the brain, researchers have already demonstrated rudimentary control of robotic prosthetic arms in primate and human trials. Now, the next generation of neural interface electronics must provide the required advances in size and power consumption to enable long-term viability of complex, high degree-of-freedom prosthetic devices.

This dissertation presents two complete neural interface systems to address two key challenges: evading the brain's foreign body response to achieve long probe longevity, and scaling wireless, implantable systems to high channel counts. The first, a self-contained, 0.125 mm2, 4-channel wireless recording system, achieves an unprecedented level of miniaturization. This opens the possibility of free-floating neural nodes in the brain tissue, which eliminates strain caused by transcranial wires. Ultimately, this may lead to probes that out- smart the brain's biological response, and provide stable, long-term recordings for chronic brain-machine interfaces. The second system achieves an unprecedented level of integration, combining 64 recording channels, 16 stimulation channels, and neural data compression onto a single 4.78 mm2 IC. Furthermore, the IC achieves substantial improvements in power and area versus state-of-the-art. These improvements in performance and functionality enable neural recording systems that scale up to thousands of channels, or scale down to extremely compact, low weight, low area, wireless interfaces.