Much of the current recording of brain waves measures signals from discrete neurons often over a few days or weeks. Medtronic's device and ones like it will take readings from "ensembles" of a few thousand neurons over periods of the seven-to-ten year average lifespan of an implant.
And they will listen very carefully.
"The signals are on the order of a micro-volt" typically in a 0-100 Hertz band, said Denison. "These signals are an order of magnitude or more lower than classic cardiac signals," he said.
Each of the two electrodes snaking off the implant has four contacts that measure 6 millimeters-squared each. That lets the device monitor four neural sites simultaneously, dissipating about five micro-watts per channel to extract the signals and process them.
"We don't have milli-watts to throw around," said Denison. "We have a budget of 10 micro-watts or less to add any sensing capabilities," he said.
If the device senses something going awry, the leads can deliver 100 micro-watts of power to stimulate neurons. "That is enough to make a meaningful difference in therapy," said Molnar.
Medtronic will make products optimized to sense patterns of specific neural disorders such as Parkinson's, epilepsy or depression. And each device will be specifically programmed to tune into the variations in brain waves of individual patients.
Denison describes the 2x2 mm chip at the heart of the device as a sort of AM radio, a spectral processor that tunes into specific bands of energy. It uses still-evolving algorithms to extract the signals from a noisy environment, amplify them and process them to make decisions about what actions to take.
The chip started its life more than two years ago as "a simple electroencephalograph amplifier, but that didn't get us very far," said Denison, so designers started adding signal processing capabilities.
The 0.8 micron chip is packaged in a stack with an SRAM of less than 10 Mbytes in capacity. The board includes a custom-built three-axis accelerometer to record data about the position of the head.
|Tim Denison of Medtronic shows the pc board of the brain implant he helped design.|
Denison and three colleagues spent the last year validating the design now being implanted in monkeys for research on brain waves. The next step will be to try the device out in humans as a way patients can control artificial limbs with their thoughts.
"We see motor prosthesis as an outstanding proving ground," said Denison. "With epilepsy and Parkinson's the science and the technology are both unknown which makes it very complex, but in the prosthesis space the science is mature and we can focus on the technology," he said.
Once the device is proven out as a way to control artificial limbs, Medtronic will turn to the scientific work of studying epilepsy and Parkinson's brain patterns to apply the device for use on those diseases. It's unclear how long all the trials and regulatory approvals will take before Medtronic can ship its first closed-loop brain stimulator.
"I would estimate we are where cardiac implants were 30-40 years ago," said Denison. "The neuroscience is evolving in real time, and we are just at the cusp making some significant discoveries in the neuroscience," he said.